feat(hermes): add tts voice controls

This commit is contained in:
晴天
2026-05-27 04:02:36 +08:00
parent 68320496b3
commit c196cef026
8 changed files with 1149 additions and 1 deletions

View File

@@ -3330,6 +3330,8 @@ const HERMES_STT_PROVIDERS = new Set(['auto', 'local', 'groq', 'openai', 'mistra
const HERMES_STT_LOCAL_MODELS = new Set(['tiny', 'base', 'small', 'medium', 'large-v3', 'turbo'])
const HERMES_STT_OPENAI_MODELS = new Set(['whisper-1', 'gpt-4o-mini-transcribe', 'gpt-4o-transcribe'])
const HERMES_STT_MISTRAL_MODELS = new Set(['voxtral-mini-latest', 'voxtral-mini-2602'])
const HERMES_TTS_PROVIDERS = new Set(['edge', 'elevenlabs', 'openai', 'xai', 'minimax', 'mistral', 'gemini', 'neutts', 'kittentts', 'piper'])
const HERMES_TTS_OPENAI_VOICES = new Set(['alloy', 'echo', 'fable', 'onyx', 'nova', 'shimmer'])
const HERMES_AUXILIARY_PROVIDERS = new Set(['auto', 'openrouter', 'nous', 'gemini', 'ollama-cloud', 'codex', 'main'])
const HERMES_APPROVAL_MODES = new Set(['manual', 'smart', 'off'])
const HERMES_APPROVAL_CRON_MODES = new Set(['deny', 'approve'])
@@ -3506,6 +3508,28 @@ function normalizeHermesSttLanguage(value, strict = false) {
return ''
}
function normalizeHermesTtsProvider(value, strict = false) {
const provider = String(value ?? '').trim().toLowerCase() || 'edge'
if (HERMES_TTS_PROVIDERS.has(provider)) return provider
if (strict) throw new Error('tts.provider 必须是 edge、elevenlabs、openai、xai、minimax、mistral、gemini、neutts、kittentts 或 piper')
return 'edge'
}
function normalizeHermesTtsOpenaiVoice(value, strict = false) {
const voice = String(value ?? '').trim().toLowerCase() || 'alloy'
if (HERMES_TTS_OPENAI_VOICES.has(voice)) return voice
if (strict) throw new Error('tts.openai.voice 必须是 alloy、echo、fable、onyx、nova 或 shimmer')
return 'alloy'
}
function normalizeHermesVoiceLanguage(value, strict = false, key = 'tts.xai.language') {
const language = String(value ?? '').trim()
if (!language) return 'en'
if (/^[a-z]{2,3}(-[A-Za-z0-9]+)?$/.test(language)) return language
if (strict) throw new Error(`${key} 必须是合法语言标签,例如 en、zh、pt-BR`)
return 'en'
}
function normalizeHermesApprovalMode(value, strict = false) {
const mode = String(value ?? '').trim().toLowerCase() || 'manual'
if (HERMES_APPROVAL_MODES.has(mode)) return mode
@@ -5479,6 +5503,92 @@ export function buildHermesSttConfigValues(config = {}) {
}
}
export function buildHermesTtsVoiceConfigValues(config = {}) {
const root = config && typeof config === 'object' && !Array.isArray(config) ? config : {}
const tts = root.tts && typeof root.tts === 'object' && !Array.isArray(root.tts) ? root.tts : {}
const edge = tts.edge && typeof tts.edge === 'object' && !Array.isArray(tts.edge) ? tts.edge : {}
const openai = tts.openai && typeof tts.openai === 'object' && !Array.isArray(tts.openai) ? tts.openai : {}
const elevenlabs = tts.elevenlabs && typeof tts.elevenlabs === 'object' && !Array.isArray(tts.elevenlabs) ? tts.elevenlabs : {}
const xai = tts.xai && typeof tts.xai === 'object' && !Array.isArray(tts.xai) ? tts.xai : {}
const mistral = tts.mistral && typeof tts.mistral === 'object' && !Array.isArray(tts.mistral) ? tts.mistral : {}
const piper = tts.piper && typeof tts.piper === 'object' && !Array.isArray(tts.piper) ? tts.piper : {}
const voice = root.voice && typeof root.voice === 'object' && !Array.isArray(root.voice) ? root.voice : {}
return {
ttsProvider: normalizeHermesTtsProvider(tts.provider, false),
ttsEdgeVoice: typeof edge.voice === 'string' ? edge.voice.trim() : 'en-US-AriaNeural',
ttsOpenaiModel: typeof openai.model === 'string' && openai.model.trim() ? openai.model.trim() : 'gpt-4o-mini-tts',
ttsOpenaiVoice: normalizeHermesTtsOpenaiVoice(openai.voice, false),
ttsElevenlabsVoiceId: typeof elevenlabs.voice_id === 'string' ? elevenlabs.voice_id.trim() : 'pNInz6obpgDQGcFmaJgB',
ttsElevenlabsModelId: typeof elevenlabs.model_id === 'string' ? elevenlabs.model_id.trim() : 'eleven_multilingual_v2',
ttsXaiVoiceId: typeof xai.voice_id === 'string' && xai.voice_id.trim() ? xai.voice_id.trim() : 'eve',
ttsXaiLanguage: normalizeHermesVoiceLanguage(xai.language, false, 'tts.xai.language'),
ttsXaiSampleRate: parseHermesInteger(xai.sample_rate, 'tts.xai.sample_rate', 24000, 8000, 192000, false),
ttsXaiBitRate: parseHermesInteger(xai.bit_rate, 'tts.xai.bit_rate', 128000, 16000, 512000, false),
ttsMistralModel: typeof mistral.model === 'string' && mistral.model.trim() ? mistral.model.trim() : 'voxtral-mini-tts-2603',
ttsMistralVoiceId: typeof mistral.voice_id === 'string' ? mistral.voice_id.trim() : 'c69964a6-ab8b-4f8a-9465-ec0925096ec8',
ttsPiperVoice: typeof piper.voice === 'string' ? piper.voice.trim() : 'en_US-lessac-medium',
voiceRecordKey: typeof voice.record_key === 'string' ? voice.record_key.trim() : 'ctrl+b',
voiceMaxRecordingSeconds: parseHermesInteger(voice.max_recording_seconds, 'voice.max_recording_seconds', 120, 1, 3600, false),
voiceAutoTts: readHermesBool(voice.auto_tts, false),
voiceBeepEnabled: readHermesBool(voice.beep_enabled, true),
voiceSilenceThreshold: parseHermesInteger(voice.silence_threshold, 'voice.silence_threshold', 200, 0, 32767, false),
voiceSilenceDuration: parseHermesFloat(voice.silence_duration, 'voice.silence_duration', 3, 0.1, 60, false),
}
}
export function mergeHermesTtsVoiceConfig(config = {}, form = {}) {
const next = mergeConfigsPreservingFields({}, config && typeof config === 'object' && !Array.isArray(config) ? config : {})
const currentValues = buildHermesTtsVoiceConfigValues(next)
const tts = next.tts && typeof next.tts === 'object' && !Array.isArray(next.tts) ? mergeConfigsPreservingFields(next.tts, {}) : {}
const edge = tts.edge && typeof tts.edge === 'object' && !Array.isArray(tts.edge) ? mergeConfigsPreservingFields(tts.edge, {}) : {}
const openai = tts.openai && typeof tts.openai === 'object' && !Array.isArray(tts.openai) ? mergeConfigsPreservingFields(tts.openai, {}) : {}
const elevenlabs = tts.elevenlabs && typeof tts.elevenlabs === 'object' && !Array.isArray(tts.elevenlabs) ? mergeConfigsPreservingFields(tts.elevenlabs, {}) : {}
const xai = tts.xai && typeof tts.xai === 'object' && !Array.isArray(tts.xai) ? mergeConfigsPreservingFields(tts.xai, {}) : {}
const mistral = tts.mistral && typeof tts.mistral === 'object' && !Array.isArray(tts.mistral) ? mergeConfigsPreservingFields(tts.mistral, {}) : {}
const piper = tts.piper && typeof tts.piper === 'object' && !Array.isArray(tts.piper) ? mergeConfigsPreservingFields(tts.piper, {}) : {}
const voice = next.voice && typeof next.voice === 'object' && !Array.isArray(next.voice) ? mergeConfigsPreservingFields(next.voice, {}) : {}
tts.provider = normalizeHermesTtsProvider(Object.hasOwn(form, 'ttsProvider') ? form.ttsProvider : currentValues.ttsProvider, true)
const edgeVoice = normalizeHermesOptionalString(Object.hasOwn(form, 'ttsEdgeVoice') ? form.ttsEdgeVoice : currentValues.ttsEdgeVoice, 'tts.edge.voice')
if (edgeVoice) edge.voice = edgeVoice
else delete edge.voice
openai.model = normalizeHermesOptionalString(Object.hasOwn(form, 'ttsOpenaiModel') ? form.ttsOpenaiModel : currentValues.ttsOpenaiModel, 'tts.openai.model') || 'gpt-4o-mini-tts'
openai.voice = normalizeHermesTtsOpenaiVoice(Object.hasOwn(form, 'ttsOpenaiVoice') ? form.ttsOpenaiVoice : currentValues.ttsOpenaiVoice, true)
const elevenlabsVoiceId = normalizeHermesOptionalString(Object.hasOwn(form, 'ttsElevenlabsVoiceId') ? form.ttsElevenlabsVoiceId : currentValues.ttsElevenlabsVoiceId, 'tts.elevenlabs.voice_id')
if (elevenlabsVoiceId) elevenlabs.voice_id = elevenlabsVoiceId
else delete elevenlabs.voice_id
const elevenlabsModelId = normalizeHermesOptionalString(Object.hasOwn(form, 'ttsElevenlabsModelId') ? form.ttsElevenlabsModelId : currentValues.ttsElevenlabsModelId, 'tts.elevenlabs.model_id')
if (elevenlabsModelId) elevenlabs.model_id = elevenlabsModelId
else delete elevenlabs.model_id
xai.voice_id = normalizeHermesOptionalString(Object.hasOwn(form, 'ttsXaiVoiceId') ? form.ttsXaiVoiceId : currentValues.ttsXaiVoiceId, 'tts.xai.voice_id') || 'eve'
xai.language = normalizeHermesVoiceLanguage(Object.hasOwn(form, 'ttsXaiLanguage') ? form.ttsXaiLanguage : currentValues.ttsXaiLanguage, true, 'tts.xai.language')
xai.sample_rate = parseHermesInteger(Object.hasOwn(form, 'ttsXaiSampleRate') ? form.ttsXaiSampleRate : currentValues.ttsXaiSampleRate, 'tts.xai.sample_rate', 24000, 8000, 192000, true)
xai.bit_rate = parseHermesInteger(Object.hasOwn(form, 'ttsXaiBitRate') ? form.ttsXaiBitRate : currentValues.ttsXaiBitRate, 'tts.xai.bit_rate', 128000, 16000, 512000, true)
mistral.model = normalizeHermesOptionalString(Object.hasOwn(form, 'ttsMistralModel') ? form.ttsMistralModel : currentValues.ttsMistralModel, 'tts.mistral.model') || 'voxtral-mini-tts-2603'
const mistralVoiceId = normalizeHermesOptionalString(Object.hasOwn(form, 'ttsMistralVoiceId') ? form.ttsMistralVoiceId : currentValues.ttsMistralVoiceId, 'tts.mistral.voice_id')
if (mistralVoiceId) mistral.voice_id = mistralVoiceId
else delete mistral.voice_id
const piperVoice = normalizeHermesOptionalString(Object.hasOwn(form, 'ttsPiperVoice') ? form.ttsPiperVoice : currentValues.ttsPiperVoice, 'tts.piper.voice')
if (piperVoice) piper.voice = piperVoice
else delete piper.voice
const recordKey = normalizeHermesOptionalString(Object.hasOwn(form, 'voiceRecordKey') ? form.voiceRecordKey : currentValues.voiceRecordKey, 'voice.record_key')
if (recordKey) voice.record_key = recordKey
else delete voice.record_key
voice.max_recording_seconds = parseHermesInteger(Object.hasOwn(form, 'voiceMaxRecordingSeconds') ? form.voiceMaxRecordingSeconds : currentValues.voiceMaxRecordingSeconds, 'voice.max_recording_seconds', 120, 1, 3600, true)
voice.auto_tts = formHermesBool(form, 'voiceAutoTts', currentValues.voiceAutoTts)
voice.beep_enabled = formHermesBool(form, 'voiceBeepEnabled', currentValues.voiceBeepEnabled)
voice.silence_threshold = parseHermesInteger(Object.hasOwn(form, 'voiceSilenceThreshold') ? form.voiceSilenceThreshold : currentValues.voiceSilenceThreshold, 'voice.silence_threshold', 200, 0, 32767, true)
voice.silence_duration = parseHermesFloat(Object.hasOwn(form, 'voiceSilenceDuration') ? form.voiceSilenceDuration : currentValues.voiceSilenceDuration, 'voice.silence_duration', 3, 0.1, 60, true)
tts.edge = edge
tts.openai = openai
tts.elevenlabs = elevenlabs
tts.xai = xai
tts.mistral = mistral
tts.piper = piper
next.tts = tts
next.voice = voice
return next
}
export function mergeHermesSttConfig(config = {}, form = {}) {
const next = mergeConfigsPreservingFields({}, config && typeof config === 'object' && !Array.isArray(config) ? config : {})
const currentValues = buildHermesSttConfigValues(next)
@@ -12668,6 +12778,27 @@ const handlers = {
}
},
hermes_tts_voice_config_read() {
const { configPath, exists, config } = readHermesConfigYamlObject()
return {
exists,
configPath,
values: buildHermesTtsVoiceConfigValues(config),
}
},
hermes_tts_voice_config_save({ form } = {}) {
const { configPath, config } = readHermesConfigYamlObject()
const next = mergeHermesTtsVoiceConfig(config, form || {})
const backup = writeHermesConfigYamlObject(configPath, next)
return {
ok: true,
configPath,
backup,
values: buildHermesTtsVoiceConfigValues(next),
}
},
hermes_terminal_config_read() {
const { configPath, exists, config } = readHermesConfigYamlObject()
return {

View File

@@ -7228,6 +7228,85 @@ fn normalize_hermes_stt_language(value: Option<String>, strict: bool) -> Result<
}
}
fn normalize_hermes_tts_provider(value: Option<String>, strict: bool) -> Result<String, String> {
let provider = value.unwrap_or_default().trim().to_ascii_lowercase();
let provider = if provider.is_empty() {
"edge".to_string()
} else {
provider
};
if matches!(
provider.as_str(),
"edge"
| "elevenlabs"
| "openai"
| "xai"
| "minimax"
| "mistral"
| "gemini"
| "neutts"
| "kittentts"
| "piper"
) {
return Ok(provider);
}
if strict {
Err("tts.provider 必须是 edge、elevenlabs、openai、xai、minimax、mistral、gemini、neutts、kittentts 或 piper".to_string())
} else {
Ok("edge".to_string())
}
}
fn normalize_hermes_tts_openai_voice(
value: Option<String>,
strict: bool,
) -> Result<String, String> {
let voice = value.unwrap_or_default().trim().to_ascii_lowercase();
let voice = if voice.is_empty() {
"alloy".to_string()
} else {
voice
};
if matches!(
voice.as_str(),
"alloy" | "echo" | "fable" | "onyx" | "nova" | "shimmer"
) {
return Ok(voice);
}
if strict {
Err("tts.openai.voice 必须是 alloy、echo、fable、onyx、nova 或 shimmer".to_string())
} else {
Ok("alloy".to_string())
}
}
fn normalize_hermes_voice_language(
value: Option<String>,
strict: bool,
key: &str,
) -> Result<String, String> {
let language = value.unwrap_or_default().trim().to_string();
if language.is_empty() {
return Ok("en".to_string());
}
let mut parts = language.split('-');
let Some(first) = parts.next() else {
return Ok("en".to_string());
};
let first_valid =
(2..=3).contains(&first.len()) && first.chars().all(|ch| ch.is_ascii_lowercase());
let rest_valid =
parts.all(|part| !part.is_empty() && part.chars().all(|ch| ch.is_ascii_alphanumeric()));
if first_valid && rest_valid {
return Ok(language);
}
if strict {
Err(format!("{key} 必须是合法语言标签,例如 en、zh、pt-BR"))
} else {
Ok("en".to_string())
}
}
fn normalize_hermes_approval_mode(value: Option<String>, strict: bool) -> Result<String, String> {
let mode = value.unwrap_or_default().trim().to_ascii_lowercase();
let mode = if mode.is_empty() {
@@ -8426,6 +8505,233 @@ fn merge_hermes_stt_config(config: &mut serde_yaml::Value, form: &Value) -> Resu
Ok(())
}
fn build_hermes_tts_voice_config_values(config: &serde_yaml::Value) -> Value {
let root = config.as_mapping();
let tts = root.and_then(|map| yaml_get_mapping(map, "tts"));
let edge = tts.and_then(|map| yaml_get_mapping(map, "edge"));
let openai = tts.and_then(|map| yaml_get_mapping(map, "openai"));
let elevenlabs = tts.and_then(|map| yaml_get_mapping(map, "elevenlabs"));
let xai = tts.and_then(|map| yaml_get_mapping(map, "xai"));
let mistral = tts.and_then(|map| yaml_get_mapping(map, "mistral"));
let piper = tts.and_then(|map| yaml_get_mapping(map, "piper"));
let voice = root.and_then(|map| yaml_get_mapping(map, "voice"));
let tts_string = |section: Option<&serde_yaml::Mapping>, key: &str, fallback: &str| {
section
.and_then(|map| yaml_string_field(map, key))
.map(|value| value.trim().to_string())
.filter(|value| !value.is_empty())
.unwrap_or_else(|| fallback.to_string())
};
serde_json::json!({
"ttsProvider": normalize_hermes_tts_provider(tts.and_then(|map| yaml_string_field(map, "provider")), false).unwrap_or_else(|_| "edge".to_string()),
"ttsEdgeVoice": tts_string(edge, "voice", "en-US-AriaNeural"),
"ttsOpenaiModel": tts_string(openai, "model", "gpt-4o-mini-tts"),
"ttsOpenaiVoice": normalize_hermes_tts_openai_voice(openai.and_then(|map| yaml_string_field(map, "voice")), false).unwrap_or_else(|_| "alloy".to_string()),
"ttsElevenlabsVoiceId": tts_string(elevenlabs, "voice_id", "pNInz6obpgDQGcFmaJgB"),
"ttsElevenlabsModelId": tts_string(elevenlabs, "model_id", "eleven_multilingual_v2"),
"ttsXaiVoiceId": tts_string(xai, "voice_id", "eve"),
"ttsXaiLanguage": normalize_hermes_voice_language(xai.and_then(|map| yaml_string_field(map, "language")), false, "tts.xai.language").unwrap_or_else(|_| "en".to_string()),
"ttsXaiSampleRate": xai.map(|map| bounded_hermes_i64(yaml_i64_field(map, "sample_rate"), 24000, 8000, 192000)).unwrap_or(24000),
"ttsXaiBitRate": xai.map(|map| bounded_hermes_i64(yaml_i64_field(map, "bit_rate"), 128000, 16000, 512000)).unwrap_or(128000),
"ttsMistralModel": tts_string(mistral, "model", "voxtral-mini-tts-2603"),
"ttsMistralVoiceId": tts_string(mistral, "voice_id", "c69964a6-ab8b-4f8a-9465-ec0925096ec8"),
"ttsPiperVoice": tts_string(piper, "voice", "en_US-lessac-medium"),
"voiceRecordKey": voice.and_then(|map| yaml_string_field(map, "record_key")).map(|value| value.trim().to_string()).unwrap_or_else(|| "ctrl+b".to_string()),
"voiceMaxRecordingSeconds": voice.map(|map| bounded_hermes_i64(yaml_i64_field(map, "max_recording_seconds"), 120, 1, 3600)).unwrap_or(120),
"voiceAutoTts": voice.and_then(|map| yaml_bool_field(map, "auto_tts")).unwrap_or(false),
"voiceBeepEnabled": voice.and_then(|map| yaml_bool_field(map, "beep_enabled")).unwrap_or(true),
"voiceSilenceThreshold": voice.map(|map| bounded_hermes_i64(yaml_i64_field(map, "silence_threshold"), 200, 0, 32767)).unwrap_or(200),
"voiceSilenceDuration": voice.map(|map| bounded_hermes_f64(yaml_f64_field(map, "silence_duration"), 3.0, 0.1, 60.0)).unwrap_or(3.0),
})
}
fn merge_hermes_tts_voice_config(
config: &mut serde_yaml::Value,
form: &Value,
) -> Result<(), String> {
let current = build_hermes_tts_voice_config_values(config);
let form_or_current_string = |key: &str| {
if form.get(key).is_some() {
form_string(form, key)
} else {
current[key].as_str().map(ToString::to_string)
}
};
let tts_provider = normalize_hermes_tts_provider(form_or_current_string("ttsProvider"), true)?;
let tts_edge_voice = form_or_current_string("ttsEdgeVoice")
.unwrap_or_default()
.trim()
.to_string();
let tts_openai_model = form_or_current_string("ttsOpenaiModel")
.map(|value| value.trim().to_string())
.filter(|value| !value.is_empty())
.unwrap_or_else(|| "gpt-4o-mini-tts".to_string());
let tts_openai_voice =
normalize_hermes_tts_openai_voice(form_or_current_string("ttsOpenaiVoice"), true)?;
let tts_elevenlabs_voice_id = form_or_current_string("ttsElevenlabsVoiceId")
.unwrap_or_default()
.trim()
.to_string();
let tts_elevenlabs_model_id = form_or_current_string("ttsElevenlabsModelId")
.unwrap_or_default()
.trim()
.to_string();
let tts_xai_voice_id = form_or_current_string("ttsXaiVoiceId")
.map(|value| value.trim().to_string())
.filter(|value| !value.is_empty())
.unwrap_or_else(|| "eve".to_string());
let tts_xai_language = normalize_hermes_voice_language(
form_or_current_string("ttsXaiLanguage"),
true,
"tts.xai.language",
)?;
let tts_xai_sample_rate = validate_hermes_i64(
if form.get("ttsXaiSampleRate").is_some() {
form_i64(form, "ttsXaiSampleRate")
} else {
current["ttsXaiSampleRate"].as_i64()
},
"tts.xai.sample_rate",
24000,
8000,
192000,
)?;
let tts_xai_bit_rate = validate_hermes_i64(
if form.get("ttsXaiBitRate").is_some() {
form_i64(form, "ttsXaiBitRate")
} else {
current["ttsXaiBitRate"].as_i64()
},
"tts.xai.bit_rate",
128000,
16000,
512000,
)?;
let tts_mistral_model = form_or_current_string("ttsMistralModel")
.map(|value| value.trim().to_string())
.filter(|value| !value.is_empty())
.unwrap_or_else(|| "voxtral-mini-tts-2603".to_string());
let tts_mistral_voice_id = form_or_current_string("ttsMistralVoiceId")
.unwrap_or_default()
.trim()
.to_string();
let tts_piper_voice = form_or_current_string("ttsPiperVoice")
.unwrap_or_default()
.trim()
.to_string();
let voice_record_key = form_or_current_string("voiceRecordKey")
.unwrap_or_default()
.trim()
.to_string();
let voice_max_recording_seconds = validate_hermes_i64(
if form.get("voiceMaxRecordingSeconds").is_some() {
form_i64(form, "voiceMaxRecordingSeconds")
} else {
current["voiceMaxRecordingSeconds"].as_i64()
},
"voice.max_recording_seconds",
120,
1,
3600,
)?;
let voice_auto_tts = form_bool(form, "voiceAutoTts")
.unwrap_or_else(|| current["voiceAutoTts"].as_bool().unwrap_or(false));
let voice_beep_enabled = form_bool(form, "voiceBeepEnabled")
.unwrap_or_else(|| current["voiceBeepEnabled"].as_bool().unwrap_or(true));
let voice_silence_threshold = validate_hermes_i64(
if form.get("voiceSilenceThreshold").is_some() {
form_i64(form, "voiceSilenceThreshold")
} else {
current["voiceSilenceThreshold"].as_i64()
},
"voice.silence_threshold",
200,
0,
32767,
)?;
let voice_silence_duration = validate_hermes_f64(
if form.get("voiceSilenceDuration").is_some() {
form_f64(form, "voiceSilenceDuration")
} else {
current["voiceSilenceDuration"].as_f64()
},
"voice.silence_duration",
3.0,
0.1,
60.0,
)?;
let root = ensure_yaml_object(config)?;
let tts = yaml_child_object(root, "tts")?;
tts.insert(
yaml_key("provider"),
serde_yaml::Value::String(tts_provider),
);
let edge = yaml_child_object(tts, "edge")?;
set_optional_yaml_string(edge, "voice", tts_edge_voice);
let openai = yaml_child_object(tts, "openai")?;
openai.insert(
yaml_key("model"),
serde_yaml::Value::String(tts_openai_model),
);
openai.insert(
yaml_key("voice"),
serde_yaml::Value::String(tts_openai_voice),
);
let elevenlabs = yaml_child_object(tts, "elevenlabs")?;
set_optional_yaml_string(elevenlabs, "voice_id", tts_elevenlabs_voice_id);
set_optional_yaml_string(elevenlabs, "model_id", tts_elevenlabs_model_id);
let xai = yaml_child_object(tts, "xai")?;
xai.insert(
yaml_key("voice_id"),
serde_yaml::Value::String(tts_xai_voice_id),
);
xai.insert(
yaml_key("language"),
serde_yaml::Value::String(tts_xai_language),
);
xai.insert(
yaml_key("sample_rate"),
serde_yaml::Value::Number(tts_xai_sample_rate.into()),
);
xai.insert(
yaml_key("bit_rate"),
serde_yaml::Value::Number(tts_xai_bit_rate.into()),
);
let mistral = yaml_child_object(tts, "mistral")?;
mistral.insert(
yaml_key("model"),
serde_yaml::Value::String(tts_mistral_model),
);
set_optional_yaml_string(mistral, "voice_id", tts_mistral_voice_id);
let piper = yaml_child_object(tts, "piper")?;
set_optional_yaml_string(piper, "voice", tts_piper_voice);
let voice = yaml_child_object(root, "voice")?;
set_optional_yaml_string(voice, "record_key", voice_record_key);
voice.insert(
yaml_key("max_recording_seconds"),
serde_yaml::Value::Number(voice_max_recording_seconds.into()),
);
voice.insert(
yaml_key("auto_tts"),
serde_yaml::Value::Bool(voice_auto_tts),
);
voice.insert(
yaml_key("beep_enabled"),
serde_yaml::Value::Bool(voice_beep_enabled),
);
voice.insert(
yaml_key("silence_threshold"),
serde_yaml::Value::Number(voice_silence_threshold.into()),
);
voice.insert(
yaml_key("silence_duration"),
serde_yaml::to_value(voice_silence_duration).map_err(|err| err.to_string())?,
);
Ok(())
}
fn merge_hermes_execution_limits_config(
config: &mut serde_yaml::Value,
form: &Value,
@@ -10688,6 +10994,30 @@ pub fn hermes_stt_config_save(form: Value) -> Result<Value, String> {
}))
}
#[tauri::command]
pub fn hermes_tts_voice_config_read() -> Result<Value, String> {
let (config_path, exists, config) = read_hermes_channel_yaml_config()?;
ensure_yaml_object(&mut config.clone())?;
Ok(serde_json::json!({
"exists": exists,
"configPath": config_path.to_string_lossy(),
"values": build_hermes_tts_voice_config_values(&config),
}))
}
#[tauri::command]
pub fn hermes_tts_voice_config_save(form: Value) -> Result<Value, String> {
let (config_path, _exists, mut config) = read_hermes_channel_yaml_config()?;
merge_hermes_tts_voice_config(&mut config, &form)?;
let backup = write_hermes_yaml_config(&config_path, &config)?;
Ok(serde_json::json!({
"ok": true,
"configPath": config_path.to_string_lossy(),
"backup": backup,
"values": build_hermes_tts_voice_config_values(&config),
}))
}
#[tauri::command]
pub fn hermes_terminal_config_read() -> Result<Value, String> {
let (config_path, exists, config) = read_hermes_channel_yaml_config()?;
@@ -17241,6 +17571,242 @@ memory:
}
}
#[cfg(test)]
mod hermes_tts_voice_config_tests {
use super::{build_hermes_tts_voice_config_values, merge_hermes_tts_voice_config};
use serde_json::json;
#[test]
fn tts_voice_values_have_upstream_defaults() {
let config: serde_yaml::Value = serde_yaml::from_str("{}").unwrap();
let values = build_hermes_tts_voice_config_values(&config);
assert_eq!(values["ttsProvider"], "edge");
assert_eq!(values["ttsEdgeVoice"], "en-US-AriaNeural");
assert_eq!(values["ttsOpenaiModel"], "gpt-4o-mini-tts");
assert_eq!(values["ttsOpenaiVoice"], "alloy");
assert_eq!(values["ttsElevenlabsVoiceId"], "pNInz6obpgDQGcFmaJgB");
assert_eq!(values["ttsElevenlabsModelId"], "eleven_multilingual_v2");
assert_eq!(values["ttsXaiVoiceId"], "eve");
assert_eq!(values["ttsXaiLanguage"], "en");
assert_eq!(values["ttsXaiSampleRate"], 24000);
assert_eq!(values["ttsXaiBitRate"], 128000);
assert_eq!(values["ttsMistralModel"], "voxtral-mini-tts-2603");
assert_eq!(
values["ttsMistralVoiceId"],
"c69964a6-ab8b-4f8a-9465-ec0925096ec8"
);
assert_eq!(values["ttsPiperVoice"], "en_US-lessac-medium");
assert_eq!(values["voiceRecordKey"], "ctrl+b");
assert_eq!(values["voiceMaxRecordingSeconds"], 120);
assert_eq!(values["voiceAutoTts"], false);
assert_eq!(values["voiceBeepEnabled"], true);
assert_eq!(values["voiceSilenceThreshold"], 200);
assert_eq!(values["voiceSilenceDuration"], 3.0);
}
#[test]
fn tts_voice_values_read_yaml_fields() {
let config: serde_yaml::Value = serde_yaml::from_str(
r#"
tts:
provider: openai
edge:
voice: zh-CN-XiaoxiaoNeural
openai:
model: gpt-4o-mini-tts
voice: nova
elevenlabs:
voice_id: voice-123
model_id: eleven_turbo_v2_5
xai:
voice_id: custom-eve
language: zh
sample_rate: 48000
bit_rate: 192000
mistral:
model: voxtral-mini-tts-2603
voice_id: mistral-voice
piper:
voice: zh_CN-huayan-medium
voice:
record_key: ctrl+shift+v
max_recording_seconds: 240
auto_tts: true
beep_enabled: false
silence_threshold: 350
silence_duration: 1.5
"#,
)
.unwrap();
let values = build_hermes_tts_voice_config_values(&config);
assert_eq!(values["ttsProvider"], "openai");
assert_eq!(values["ttsEdgeVoice"], "zh-CN-XiaoxiaoNeural");
assert_eq!(values["ttsOpenaiVoice"], "nova");
assert_eq!(values["ttsElevenlabsVoiceId"], "voice-123");
assert_eq!(values["ttsXaiLanguage"], "zh");
assert_eq!(values["ttsXaiSampleRate"], 48000);
assert_eq!(values["ttsMistralVoiceId"], "mistral-voice");
assert_eq!(values["ttsPiperVoice"], "zh_CN-huayan-medium");
assert_eq!(values["voiceRecordKey"], "ctrl+shift+v");
assert_eq!(values["voiceAutoTts"], true);
assert_eq!(values["voiceBeepEnabled"], false);
assert_eq!(values["voiceSilenceDuration"], 1.5);
}
#[test]
fn merge_tts_voice_config_preserves_unknown_fields() {
let mut config: serde_yaml::Value = serde_yaml::from_str(
r#"
model:
provider: anthropic
tts:
provider: edge
custom_flag: keep-tts
openai:
custom_flag: keep-openai
piper:
voices_dir: /cache/piper
voice:
custom_flag: keep-voice
streaming:
enabled: true
"#,
)
.unwrap();
merge_hermes_tts_voice_config(
&mut config,
&json!({
"ttsProvider": "openai",
"ttsEdgeVoice": "zh-CN-XiaoxiaoNeural",
"ttsOpenaiModel": "gpt-4o-mini-tts",
"ttsOpenaiVoice": "nova",
"ttsElevenlabsVoiceId": "voice-123",
"ttsElevenlabsModelId": "eleven_turbo_v2_5",
"ttsXaiVoiceId": "eve-pro",
"ttsXaiLanguage": "zh",
"ttsXaiSampleRate": "48000",
"ttsXaiBitRate": "192000",
"ttsMistralModel": "voxtral-mini-tts-2603",
"ttsMistralVoiceId": "mistral-voice",
"ttsPiperVoice": "zh_CN-huayan-medium",
"voiceRecordKey": "ctrl+shift+v",
"voiceMaxRecordingSeconds": "240",
"voiceAutoTts": true,
"voiceBeepEnabled": false,
"voiceSilenceThreshold": "350",
"voiceSilenceDuration": "1.5",
}),
)
.unwrap();
assert_eq!(config["model"]["provider"].as_str(), Some("anthropic"));
assert_eq!(config["streaming"]["enabled"].as_bool(), Some(true));
assert_eq!(config["tts"]["provider"].as_str(), Some("openai"));
assert_eq!(
config["tts"]["edge"]["voice"].as_str(),
Some("zh-CN-XiaoxiaoNeural")
);
assert_eq!(config["tts"]["openai"]["voice"].as_str(), Some("nova"));
assert_eq!(
config["tts"]["openai"]["custom_flag"].as_str(),
Some("keep-openai")
);
assert_eq!(
config["tts"]["elevenlabs"]["voice_id"].as_str(),
Some("voice-123")
);
assert_eq!(config["tts"]["xai"]["sample_rate"].as_i64(), Some(48000));
assert_eq!(config["tts"]["xai"]["bit_rate"].as_i64(), Some(192000));
assert_eq!(
config["tts"]["mistral"]["voice_id"].as_str(),
Some("mistral-voice")
);
assert_eq!(
config["tts"]["piper"]["voice"].as_str(),
Some("zh_CN-huayan-medium")
);
assert_eq!(
config["tts"]["piper"]["voices_dir"].as_str(),
Some("/cache/piper")
);
assert_eq!(config["tts"]["custom_flag"].as_str(), Some("keep-tts"));
assert_eq!(config["voice"]["record_key"].as_str(), Some("ctrl+shift+v"));
assert_eq!(config["voice"]["max_recording_seconds"].as_i64(), Some(240));
assert_eq!(config["voice"]["auto_tts"].as_bool(), Some(true));
assert_eq!(config["voice"]["beep_enabled"].as_bool(), Some(false));
assert_eq!(config["voice"]["silence_threshold"].as_i64(), Some(350));
assert_eq!(config["voice"]["silence_duration"].as_f64(), Some(1.5));
assert_eq!(config["voice"]["custom_flag"].as_str(), Some("keep-voice"));
}
#[test]
fn merge_tts_voice_config_removes_empty_optional_overrides() {
let mut config: serde_yaml::Value = serde_yaml::from_str(
r#"
tts:
edge:
voice: custom-edge
elevenlabs:
voice_id: voice-123
model_id: model-123
piper:
voice: custom-piper
voices_dir: /cache/piper
voice:
record_key: ctrl+shift+v
custom_flag: keep-voice
"#,
)
.unwrap();
merge_hermes_tts_voice_config(
&mut config,
&json!({
"ttsEdgeVoice": "",
"ttsElevenlabsVoiceId": " ",
"ttsElevenlabsModelId": "",
"ttsPiperVoice": "",
"voiceRecordKey": "",
}),
)
.unwrap();
assert!(config["tts"]["edge"]["voice"].is_null());
assert!(config["tts"]["elevenlabs"]["voice_id"].is_null());
assert!(config["tts"]["elevenlabs"]["model_id"].is_null());
assert!(config["tts"]["piper"]["voice"].is_null());
assert_eq!(
config["tts"]["piper"]["voices_dir"].as_str(),
Some("/cache/piper")
);
assert!(config["voice"]["record_key"].is_null());
assert_eq!(config["voice"]["custom_flag"].as_str(), Some("keep-voice"));
}
#[test]
fn merge_tts_voice_config_rejects_invalid_values() {
let mut config = serde_yaml::Value::Mapping(serde_yaml::Mapping::new());
let err = merge_hermes_tts_voice_config(&mut config, &json!({ "ttsProvider": "bad" }))
.unwrap_err();
assert!(err.contains("tts.provider"));
let err = merge_hermes_tts_voice_config(&mut config, &json!({ "ttsOpenaiVoice": "robot" }))
.unwrap_err();
assert!(err.contains("tts.openai.voice"));
let err = merge_hermes_tts_voice_config(&mut config, &json!({ "ttsXaiSampleRate": "0" }))
.unwrap_err();
assert!(err.contains("tts.xai.sample_rate"));
let err =
merge_hermes_tts_voice_config(&mut config, &json!({ "voiceMaxRecordingSeconds": "0" }))
.unwrap_err();
assert!(err.contains("voice.max_recording_seconds"));
let err =
merge_hermes_tts_voice_config(&mut config, &json!({ "voiceSilenceDuration": "-1" }))
.unwrap_err();
assert!(err.contains("voice.silence_duration"));
}
}
#[cfg(test)]
mod hermes_checkpoints_config_tests {
use super::{build_hermes_checkpoints_config_values, merge_hermes_checkpoints_config};

View File

@@ -329,6 +329,8 @@ pub fn run() {
hermes::hermes_browser_config_save,
hermes::hermes_stt_config_read,
hermes::hermes_stt_config_save,
hermes::hermes_tts_voice_config_read,
hermes::hermes_tts_voice_config_save,
hermes::hermes_terminal_config_read,
hermes::hermes_terminal_config_save,
hermes::hermes_lazy_deps_features,

View File

@@ -303,6 +303,28 @@ const STT_DEFAULTS = {
sttMistralModel: 'voxtral-mini-latest',
}
const TTS_VOICE_DEFAULTS = {
ttsProvider: 'edge',
ttsEdgeVoice: 'en-US-AriaNeural',
ttsOpenaiModel: 'gpt-4o-mini-tts',
ttsOpenaiVoice: 'alloy',
ttsElevenlabsVoiceId: 'pNInz6obpgDQGcFmaJgB',
ttsElevenlabsModelId: 'eleven_multilingual_v2',
ttsXaiVoiceId: 'eve',
ttsXaiLanguage: 'en',
ttsXaiSampleRate: 24000,
ttsXaiBitRate: 128000,
ttsMistralModel: 'voxtral-mini-tts-2603',
ttsMistralVoiceId: 'c69964a6-ab8b-4f8a-9465-ec0925096ec8',
ttsPiperVoice: 'en_US-lessac-medium',
voiceRecordKey: 'ctrl+b',
voiceMaxRecordingSeconds: 120,
voiceAutoTts: false,
voiceBeepEnabled: true,
voiceSilenceThreshold: 200,
voiceSilenceDuration: 3,
}
const TERMINAL_DEFAULTS = {
terminalBackend: 'local',
terminalCwd: '.',
@@ -339,6 +361,8 @@ const STT_PROVIDERS = ['auto', 'local', 'groq', 'openai', 'mistral']
const STT_LOCAL_MODELS = ['tiny', 'base', 'small', 'medium', 'large-v3', 'turbo']
const STT_OPENAI_MODELS = ['whisper-1', 'gpt-4o-mini-transcribe', 'gpt-4o-transcribe']
const STT_MISTRAL_MODELS = ['voxtral-mini-latest', 'voxtral-mini-2602']
const TTS_PROVIDERS = ['edge', 'elevenlabs', 'openai', 'xai', 'minimax', 'mistral', 'gemini', 'neutts', 'kittentts', 'piper']
const TTS_OPENAI_VOICES = ['alloy', 'echo', 'fable', 'onyx', 'nova', 'shimmer']
const UNAUTHORIZED_DM_BEHAVIORS = ['pair', 'ignore']
const IMAGE_INPUT_MODES = ['auto', 'native', 'text']
const REASONING_EFFORTS = ['xhigh', 'high', 'medium', 'low', 'minimal', 'none']
@@ -399,6 +423,7 @@ export function render() {
let privacyValues = { ...PRIVACY_DEFAULTS }
let browserValues = { ...BROWSER_DEFAULTS }
let sttValues = { ...STT_DEFAULTS }
let ttsVoiceValues = { ...TTS_VOICE_DEFAULTS }
let terminalValues = { ...TERMINAL_DEFAULTS }
let loading = true
let runtimeLoading = true
@@ -437,6 +462,7 @@ export function render() {
let privacyLoading = true
let browserLoading = true
let sttLoading = true
let ttsVoiceLoading = true
let terminalLoading = true
let saving = false
let runtimeSaving = false
@@ -475,6 +501,7 @@ export function render() {
let privacySaving = false
let browserSaving = false
let sttSaving = false
let ttsVoiceSaving = false
let terminalSaving = false
let error = null
let runtimeError = null
@@ -513,6 +540,7 @@ export function render() {
let privacyError = null
let browserError = null
let sttError = null
let ttsVoiceError = null
let terminalError = null
function esc(value) {
@@ -524,7 +552,7 @@ export function render() {
}
function isBusy() {
return loading || runtimeLoading || sessionsMaintenanceLoading || updatesLoading || compressionLoading || promptCachingLoading || openrouterCacheLoading || providerRoutingLoading || auxiliaryLoading || toolGuardrailsLoading || memoryLoading || skillsLoading || curatorLoading || quickCommandsLoading || modelLoading || modelAliasesLoading || hooksLoading || providerOverridesLoading || mcpServersLoading || agentToolsetsLoading || platformToolsetsLoading || agentRuntimeLoading || unauthorizedDmLoading || securityLoading || displayLoading || humanDelayLoading || kanbanLoading || streamingLoading || executionLimitsLoading || ioSafetyLoading || checkpointsLoading || cronLoading || loggingLoading || approvalsLoading || privacyLoading || browserLoading || sttLoading || terminalLoading || saving || runtimeSaving || sessionsMaintenanceSaving || updatesSaving || compressionSaving || promptCachingSaving || openrouterCacheSaving || providerRoutingSaving || auxiliarySaving || toolGuardrailsSaving || memorySaving || skillsSaving || curatorSaving || quickCommandsSaving || modelSaving || modelAliasesSaving || hooksSaving || providerOverridesSaving || mcpServersSaving || agentToolsetsSaving || platformToolsetsSaving || agentRuntimeSaving || unauthorizedDmSaving || securitySaving || displaySaving || humanDelaySaving || kanbanSaving || streamingSaving || executionLimitsSaving || ioSafetySaving || checkpointsSaving || cronSaving || loggingSaving || approvalsSaving || privacySaving || browserSaving || sttSaving || terminalSaving
return loading || runtimeLoading || sessionsMaintenanceLoading || updatesLoading || compressionLoading || promptCachingLoading || openrouterCacheLoading || providerRoutingLoading || auxiliaryLoading || toolGuardrailsLoading || memoryLoading || skillsLoading || curatorLoading || quickCommandsLoading || modelLoading || modelAliasesLoading || hooksLoading || providerOverridesLoading || mcpServersLoading || agentToolsetsLoading || platformToolsetsLoading || agentRuntimeLoading || unauthorizedDmLoading || securityLoading || displayLoading || humanDelayLoading || kanbanLoading || streamingLoading || executionLimitsLoading || ioSafetyLoading || checkpointsLoading || cronLoading || loggingLoading || approvalsLoading || privacyLoading || browserLoading || sttLoading || ttsVoiceLoading || terminalLoading || saving || runtimeSaving || sessionsMaintenanceSaving || updatesSaving || compressionSaving || promptCachingSaving || openrouterCacheSaving || providerRoutingSaving || auxiliarySaving || toolGuardrailsSaving || memorySaving || skillsSaving || curatorSaving || quickCommandsSaving || modelSaving || modelAliasesSaving || hooksSaving || providerOverridesSaving || mcpServersSaving || agentToolsetsSaving || platformToolsetsSaving || agentRuntimeSaving || unauthorizedDmSaving || securitySaving || displaySaving || humanDelaySaving || kanbanSaving || streamingSaving || executionLimitsSaving || ioSafetySaving || checkpointsSaving || cronSaving || loggingSaving || approvalsSaving || privacySaving || browserSaving || sttSaving || ttsVoiceSaving || terminalSaving
}
function option(labelKey, value, selected) {
@@ -2271,6 +2299,112 @@ export function render() {
`
}
function renderTtsVoicePanel() {
const disabled = loading || saving || ttsVoiceLoading || ttsVoiceSaving || sttSaving || approvalsSaving || cronSaving || loggingSaving || privacySaving || browserSaving || terminalSaving || runtimeSaving || compressionSaving || promptCachingSaving || openrouterCacheSaving || providerRoutingSaving || auxiliarySaving || toolGuardrailsSaving || memorySaving || skillsSaving || quickCommandsSaving || providerOverridesSaving || agentToolsetsSaving || agentRuntimeSaving || unauthorizedDmSaving || streamingSaving || executionLimitsSaving || ioSafetySaving || checkpointsSaving
return `
<div class="hm-panel hm-config-runtime-panel hm-config-tts-voice-panel">
<div class="hm-panel-header">
<div>
<div class="hm-panel-title">${t('engine.hermesTtsVoiceConfigTitle')}</div>
<div class="hm-channel-panel-desc">${t('engine.hermesTtsVoiceConfigDesc')}</div>
</div>
<div class="hm-panel-actions">
<span class="hm-muted">${ttsVoiceSaving ? t('engine.hermesConfigStatusSaving') : ttsVoiceLoading ? t('engine.hermesConfigStatusLoading') : t('engine.hermesTtsVoiceConfigStatusReady')}</span>
<button class="hm-btn hm-btn--cta hm-btn--sm" id="hm-tts-voice-save" ${disabled ? 'disabled' : ''}>${t('engine.hermesTtsVoiceConfigSave')}</button>
</div>
</div>
<div class="hm-panel-body">
${renderError(ttsVoiceError)}
<div class="hm-config-check-grid">
<label class="hm-channel-check">
<input id="hm-voice-auto-tts" type="checkbox" ${ttsVoiceValues.voiceAutoTts ? 'checked' : ''} ${disabled ? 'disabled' : ''}>
<span>${t('engine.hermesVoiceConfigAutoTts')}</span>
</label>
<label class="hm-channel-check">
<input id="hm-voice-beep-enabled" type="checkbox" ${ttsVoiceValues.voiceBeepEnabled ? 'checked' : ''} ${disabled ? 'disabled' : ''}>
<span>${t('engine.hermesVoiceConfigBeepEnabled')}</span>
</label>
</div>
<div class="hm-config-runtime-grid hm-config-tts-grid">
<label class="hm-field">
<span class="hm-field-label">${t('engine.hermesTtsConfigProvider')}</span>
<select id="hm-tts-provider" class="hm-input" ${disabled ? 'disabled' : ''}>
${TTS_PROVIDERS.map(provider => option(`engine.hermesTtsConfigProvider_${provider}`, provider, ttsVoiceValues.ttsProvider)).join('')}
</select>
</label>
<label class="hm-field">
<span class="hm-field-label">${t('engine.hermesTtsConfigEdgeVoice')}</span>
<input id="hm-tts-edge-voice" class="hm-input" placeholder="en-US-AriaNeural" value="${esc(ttsVoiceValues.ttsEdgeVoice)}" ${disabled ? 'disabled' : ''}>
</label>
<label class="hm-field">
<span class="hm-field-label">${t('engine.hermesTtsConfigOpenaiModel')}</span>
<input id="hm-tts-openai-model" class="hm-input" placeholder="gpt-4o-mini-tts" value="${esc(ttsVoiceValues.ttsOpenaiModel)}" ${disabled ? 'disabled' : ''}>
</label>
<label class="hm-field">
<span class="hm-field-label">${t('engine.hermesTtsConfigOpenaiVoice')}</span>
<select id="hm-tts-openai-voice" class="hm-input" ${disabled ? 'disabled' : ''}>
${TTS_OPENAI_VOICES.map(voice => option(`engine.hermesTtsConfigOpenaiVoice_${voice}`, voice, ttsVoiceValues.ttsOpenaiVoice)).join('')}
</select>
</label>
<label class="hm-field">
<span class="hm-field-label">${t('engine.hermesTtsConfigElevenlabsVoiceId')}</span>
<input id="hm-tts-elevenlabs-voice-id" class="hm-input" placeholder="pNInz6obpgDQGcFmaJgB" value="${esc(ttsVoiceValues.ttsElevenlabsVoiceId)}" ${disabled ? 'disabled' : ''}>
</label>
<label class="hm-field">
<span class="hm-field-label">${t('engine.hermesTtsConfigElevenlabsModelId')}</span>
<input id="hm-tts-elevenlabs-model-id" class="hm-input" placeholder="eleven_multilingual_v2" value="${esc(ttsVoiceValues.ttsElevenlabsModelId)}" ${disabled ? 'disabled' : ''}>
</label>
<label class="hm-field">
<span class="hm-field-label">${t('engine.hermesTtsConfigXaiVoiceId')}</span>
<input id="hm-tts-xai-voice-id" class="hm-input" placeholder="eve" value="${esc(ttsVoiceValues.ttsXaiVoiceId)}" ${disabled ? 'disabled' : ''}>
</label>
<label class="hm-field">
<span class="hm-field-label">${t('engine.hermesTtsConfigXaiLanguage')}</span>
<input id="hm-tts-xai-language" class="hm-input" placeholder="en" value="${esc(ttsVoiceValues.ttsXaiLanguage)}" ${disabled ? 'disabled' : ''}>
</label>
<label class="hm-field">
<span class="hm-field-label">${t('engine.hermesTtsConfigXaiSampleRate')}</span>
<input id="hm-tts-xai-sample-rate" class="hm-input" type="number" inputmode="numeric" min="8000" max="192000" step="1000" value="${esc(ttsVoiceValues.ttsXaiSampleRate)}" ${disabled ? 'disabled' : ''}>
</label>
<label class="hm-field">
<span class="hm-field-label">${t('engine.hermesTtsConfigXaiBitRate')}</span>
<input id="hm-tts-xai-bit-rate" class="hm-input" type="number" inputmode="numeric" min="16000" max="512000" step="1000" value="${esc(ttsVoiceValues.ttsXaiBitRate)}" ${disabled ? 'disabled' : ''}>
</label>
<label class="hm-field">
<span class="hm-field-label">${t('engine.hermesTtsConfigMistralModel')}</span>
<input id="hm-tts-mistral-model" class="hm-input" placeholder="voxtral-mini-tts-2603" value="${esc(ttsVoiceValues.ttsMistralModel)}" ${disabled ? 'disabled' : ''}>
</label>
<label class="hm-field">
<span class="hm-field-label">${t('engine.hermesTtsConfigMistralVoiceId')}</span>
<input id="hm-tts-mistral-voice-id" class="hm-input" placeholder="c69964a6-ab8b-4f8a-9465-ec0925096ec8" value="${esc(ttsVoiceValues.ttsMistralVoiceId)}" ${disabled ? 'disabled' : ''}>
</label>
<label class="hm-field">
<span class="hm-field-label">${t('engine.hermesTtsConfigPiperVoice')}</span>
<input id="hm-tts-piper-voice" class="hm-input" placeholder="en_US-lessac-medium" value="${esc(ttsVoiceValues.ttsPiperVoice)}" ${disabled ? 'disabled' : ''}>
</label>
<label class="hm-field">
<span class="hm-field-label">${t('engine.hermesVoiceConfigRecordKey')}</span>
<input id="hm-voice-record-key" class="hm-input" placeholder="ctrl+b" value="${esc(ttsVoiceValues.voiceRecordKey)}" ${disabled ? 'disabled' : ''}>
</label>
<label class="hm-field">
<span class="hm-field-label">${t('engine.hermesVoiceConfigMaxRecordingSeconds')}</span>
<input id="hm-voice-max-recording-seconds" class="hm-input" type="number" inputmode="numeric" min="1" max="3600" step="1" value="${esc(ttsVoiceValues.voiceMaxRecordingSeconds)}" ${disabled ? 'disabled' : ''}>
</label>
<label class="hm-field">
<span class="hm-field-label">${t('engine.hermesVoiceConfigSilenceThreshold')}</span>
<input id="hm-voice-silence-threshold" class="hm-input" type="number" inputmode="numeric" min="0" max="32767" step="1" value="${esc(ttsVoiceValues.voiceSilenceThreshold)}" ${disabled ? 'disabled' : ''}>
</label>
<label class="hm-field">
<span class="hm-field-label">${t('engine.hermesVoiceConfigSilenceDuration')}</span>
<input id="hm-voice-silence-duration" class="hm-input" type="number" inputmode="decimal" min="0.1" max="60" step="0.1" value="${esc(ttsVoiceValues.voiceSilenceDuration)}" ${disabled ? 'disabled' : ''}>
</label>
</div>
<div class="hm-channel-footnote">${t('engine.hermesTtsVoiceConfigFootnote')}</div>
</div>
</div>
`
}
function renderTerminalPanel() {
const disabled = loading || saving || terminalLoading || terminalSaving || approvalsSaving || cronSaving || loggingSaving || browserSaving || sttSaving || runtimeSaving || compressionSaving || promptCachingSaving || openrouterCacheSaving || providerRoutingSaving || auxiliarySaving || toolGuardrailsSaving || memorySaving || skillsSaving || quickCommandsSaving || providerOverridesSaving || agentToolsetsSaving || agentRuntimeSaving || unauthorizedDmSaving || streamingSaving || executionLimitsSaving || checkpointsSaving
return `
@@ -2425,6 +2559,7 @@ export function render() {
${renderPrivacyPanel()}
${renderBrowserPanel()}
${renderSttPanel()}
${renderTtsVoicePanel()}
${renderCompressionPanel()}
${renderPromptCachingPanel()}
${renderOpenrouterCachePanel()}
@@ -2503,6 +2638,7 @@ export function render() {
el.querySelector('#hm-privacy-save')?.addEventListener('click', savePrivacyConfig)
el.querySelector('#hm-browser-save')?.addEventListener('click', saveBrowserConfig)
el.querySelector('#hm-stt-save')?.addEventListener('click', saveSttConfig)
el.querySelector('#hm-tts-voice-save')?.addEventListener('click', saveTtsVoiceConfig)
el.querySelector('#hm-terminal-save')?.addEventListener('click', saveTerminal)
}
@@ -2691,6 +2827,11 @@ export function render() {
sttValues = { ...STT_DEFAULTS, ...(data?.values || {}) }
}
async function loadTtsVoiceConfig() {
const data = await api.hermesTtsVoiceConfigRead()
ttsVoiceValues = { ...TTS_VOICE_DEFAULTS, ...(data?.values || {}) }
}
async function loadTerminal() {
const data = await api.hermesTerminalConfigRead()
terminalValues = { ...TERMINAL_DEFAULTS, ...(data?.values || {}) }
@@ -2734,6 +2875,7 @@ export function render() {
privacyLoading = true
browserLoading = true
sttLoading = true
ttsVoiceLoading = true
terminalLoading = true
error = null
runtimeError = null
@@ -2771,6 +2913,7 @@ export function render() {
privacyError = null
browserError = null
sttError = null
ttsVoiceError = null
terminalError = null
draw()
try {
@@ -2932,6 +3075,14 @@ export function render() {
sttLoading = false
draw()
}
try {
await loadTtsVoiceConfig()
} catch (err) {
ttsVoiceError = humanizeError(err, t('engine.hermesTtsVoiceConfigLoadFailed') || 'Load speech output config failed')
} finally {
ttsVoiceLoading = false
draw()
}
try {
await loadTerminal()
} catch (err) {
@@ -4264,6 +4415,49 @@ export function render() {
}
}
async function saveTtsVoiceConfig() {
const form = {
ttsProvider: el.querySelector('#hm-tts-provider')?.value || 'edge',
ttsEdgeVoice: el.querySelector('#hm-tts-edge-voice')?.value || '',
ttsOpenaiModel: el.querySelector('#hm-tts-openai-model')?.value || 'gpt-4o-mini-tts',
ttsOpenaiVoice: el.querySelector('#hm-tts-openai-voice')?.value || 'alloy',
ttsElevenlabsVoiceId: el.querySelector('#hm-tts-elevenlabs-voice-id')?.value || '',
ttsElevenlabsModelId: el.querySelector('#hm-tts-elevenlabs-model-id')?.value || '',
ttsXaiVoiceId: el.querySelector('#hm-tts-xai-voice-id')?.value || 'eve',
ttsXaiLanguage: el.querySelector('#hm-tts-xai-language')?.value || 'en',
ttsXaiSampleRate: el.querySelector('#hm-tts-xai-sample-rate')?.value || '24000',
ttsXaiBitRate: el.querySelector('#hm-tts-xai-bit-rate')?.value || '128000',
ttsMistralModel: el.querySelector('#hm-tts-mistral-model')?.value || 'voxtral-mini-tts-2603',
ttsMistralVoiceId: el.querySelector('#hm-tts-mistral-voice-id')?.value || '',
ttsPiperVoice: el.querySelector('#hm-tts-piper-voice')?.value || '',
voiceRecordKey: el.querySelector('#hm-voice-record-key')?.value || '',
voiceMaxRecordingSeconds: el.querySelector('#hm-voice-max-recording-seconds')?.value || '120',
voiceAutoTts: !!el.querySelector('#hm-voice-auto-tts')?.checked,
voiceBeepEnabled: !!el.querySelector('#hm-voice-beep-enabled')?.checked,
voiceSilenceThreshold: el.querySelector('#hm-voice-silence-threshold')?.value || '200',
voiceSilenceDuration: el.querySelector('#hm-voice-silence-duration')?.value || '3',
}
ttsVoiceSaving = true
ttsVoiceError = null
draw()
try {
const result = await api.hermesTtsVoiceConfigSave(form)
ttsVoiceValues = { ...TTS_VOICE_DEFAULTS, ...(result?.values || form) }
await refreshRawAfterStructuredSave()
const backup = result?.backup || ''
toast({
message: t('engine.hermesTtsVoiceConfigSaveSuccess'),
hint: backup ? t('engine.hermesConfigBackupHint', { path: backup }) : '',
}, 'success')
} catch (err) {
ttsVoiceError = humanizeError(err, t('engine.hermesTtsVoiceConfigSaveFailed') || 'Save speech output config failed')
toast(ttsVoiceError, 'error')
} finally {
ttsVoiceSaving = false
draw()
}
}
async function saveTerminal() {
const form = {
terminalBackend: el.querySelector('#hm-terminal-backend')?.value || 'local',

View File

@@ -581,6 +581,8 @@ export const api = {
hermesBrowserConfigSave: (form) => invoke('hermes_browser_config_save', { form }),
hermesSttConfigRead: () => invoke('hermes_stt_config_read'),
hermesSttConfigSave: (form) => invoke('hermes_stt_config_save', { form }),
hermesTtsVoiceConfigRead: () => invoke('hermes_tts_voice_config_read'),
hermesTtsVoiceConfigSave: (form) => invoke('hermes_tts_voice_config_save', { form }),
hermesTerminalConfigRead: () => invoke('hermes_terminal_config_read'),
hermesTerminalConfigSave: (form) => invoke('hermes_terminal_config_save', { form }),
hermesLazyDepsFeatures: () => cachedInvoke('hermes_lazy_deps_features', {}, 600000),

View File

@@ -739,6 +739,49 @@ export default {
'hermesSttConfigMistralModel_voxtral-mini-latest': _('voxtral-mini-latest推荐', 'voxtral-mini-latest (recommended)', 'voxtral-mini-latest建議'),
'hermesSttConfigMistralModel_voxtral-mini-2602': _('voxtral-mini-2602固定版本', 'voxtral-mini-2602 (pinned version)', 'voxtral-mini-2602固定版本'),
hermesSttConfigFootnote: _('这里写入 stt.*。API Key 仍通过 .env 管理Groq 使用上游默认模型,其他 provider 高级字段会保留在 raw YAML 中。', 'This writes stt.*. API keys are still managed through .env. Groq uses the upstream default model, and other provider advanced fields stay in raw YAML.', '這裡寫入 stt.*。API Key 仍透過 .env 管理Groq 使用上游預設模型,其他 provider 進階欄位會保留在 raw YAML 中。'),
hermesTtsVoiceConfigTitle: _('语音输出与录音', 'Speech output and recording', '語音輸出與錄音'),
hermesTtsVoiceConfigDesc: _('控制 Hermes 的 TTS 朗读 provider、主流语音模型以及 CLI 语音录制热键和静音自动停止。', 'Control Hermes TTS providers and common voice models, plus CLI recording hotkey and silence auto-stop.', '控制 Hermes 的 TTS 朗讀 provider、主流語音模型以及 CLI 語音錄製熱鍵和靜音自動停止。'),
hermesTtsVoiceConfigStatusReady: _('结构化配置', 'structured settings', '結構化設定'),
hermesTtsVoiceConfigSave: _('保存语音输出配置', 'Save speech output settings', '儲存語音輸出設定'),
hermesTtsVoiceConfigSaveSuccess: _('语音输出配置已保存,建议重启 Hermes Gateway 或 CLI 会话生效', 'Speech output settings saved. Restart Hermes Gateway or CLI sessions to take effect.', '語音輸出設定已儲存,建議重啟 Hermes Gateway 或 CLI 工作階段生效'),
hermesTtsVoiceConfigLoadFailed: _('加载语音输出配置失败', 'Load speech output settings failed', '載入語音輸出設定失敗'),
hermesTtsVoiceConfigSaveFailed: _('保存语音输出配置失败', 'Save speech output settings failed', '儲存語音輸出設定失敗'),
hermesTtsConfigProvider: _('TTS 服务', 'TTS provider', 'TTS 服務'),
hermesTtsConfigProvider_edge: _('Edge TTS免费', 'Edge TTS (free)', 'Edge TTS免費'),
hermesTtsConfigProvider_elevenlabs: _('ElevenLabs高质量', 'ElevenLabs (high quality)', 'ElevenLabs高品質'),
hermesTtsConfigProvider_openai: _('OpenAI TTS', 'OpenAI TTS', 'OpenAI TTS'),
hermesTtsConfigProvider_xai: _('xAI TTS', 'xAI TTS', 'xAI TTS'),
hermesTtsConfigProvider_minimax: _('MiniMax TTS', 'MiniMax TTS', 'MiniMax TTS'),
hermesTtsConfigProvider_mistral: _('Mistral Voxtral TTS', 'Mistral Voxtral TTS', 'Mistral Voxtral TTS'),
hermesTtsConfigProvider_gemini: _('Gemini TTS', 'Gemini TTS', 'Gemini TTS'),
hermesTtsConfigProvider_neutts: _('NeuTTS本地', 'NeuTTS (local)', 'NeuTTS本機'),
hermesTtsConfigProvider_kittentts: _('KittenTTS本地', 'KittenTTS (local)', 'KittenTTS本機'),
hermesTtsConfigProvider_piper: _('Piper本地', 'Piper (local)', 'Piper本機'),
hermesTtsConfigEdgeVoice: _('Edge 声音', 'Edge voice', 'Edge 聲音'),
hermesTtsConfigOpenaiModel: _('OpenAI 模型', 'OpenAI model', 'OpenAI 模型'),
hermesTtsConfigOpenaiVoice: _('OpenAI 声音', 'OpenAI voice', 'OpenAI 聲音'),
hermesTtsConfigOpenaiVoice_alloy: _('alloy', 'alloy', 'alloy'),
hermesTtsConfigOpenaiVoice_echo: _('echo', 'echo', 'echo'),
hermesTtsConfigOpenaiVoice_fable: _('fable', 'fable', 'fable'),
hermesTtsConfigOpenaiVoice_onyx: _('onyx', 'onyx', 'onyx'),
hermesTtsConfigOpenaiVoice_nova: _('nova', 'nova', 'nova'),
hermesTtsConfigOpenaiVoice_shimmer: _('shimmer', 'shimmer', 'shimmer'),
hermesTtsConfigElevenlabsVoiceId: _('ElevenLabs Voice ID', 'ElevenLabs voice ID', 'ElevenLabs Voice ID'),
hermesTtsConfigElevenlabsModelId: _('ElevenLabs 模型 ID', 'ElevenLabs model ID', 'ElevenLabs 模型 ID'),
hermesTtsConfigXaiVoiceId: _('xAI Voice ID', 'xAI voice ID', 'xAI Voice ID'),
hermesTtsConfigXaiLanguage: _('xAI 语言', 'xAI language', 'xAI 語言'),
hermesTtsConfigXaiSampleRate: _('xAI 采样率', 'xAI sample rate', 'xAI 取樣率'),
hermesTtsConfigXaiBitRate: _('xAI 比特率', 'xAI bit rate', 'xAI 位元率'),
hermesTtsConfigMistralModel: _('Mistral 模型', 'Mistral model', 'Mistral 模型'),
hermesTtsConfigMistralVoiceId: _('Mistral Voice ID', 'Mistral voice ID', 'Mistral Voice ID'),
hermesTtsConfigPiperVoice: _('Piper 声音', 'Piper voice', 'Piper 聲音'),
hermesVoiceConfigRecordKey: _('录音热键', 'Recording hotkey', '錄音快捷鍵'),
hermesVoiceConfigMaxRecordingSeconds: _('最长录音秒数', 'Max recording seconds', '最長錄音秒數'),
hermesVoiceConfigAutoTts: _('CLI 回复后自动朗读', 'Auto-speak CLI replies', 'CLI 回覆後自動朗讀'),
hermesVoiceConfigBeepEnabled: _('录音开始/结束提示音', 'Recording start/stop beeps', '錄音開始/結束提示音'),
hermesVoiceConfigSilenceThreshold: _('静音阈值 RMS', 'Silence RMS threshold', '靜音閾值 RMS'),
hermesVoiceConfigSilenceDuration: _('静音自动停止秒数', 'Silence auto-stop seconds', '靜音自動停止秒數'),
hermesTtsVoiceConfigFootnote: _('这里写入 tts.* 与 voice.*。API Key 仍通过 .env 管理;留空的可选声音 ID 会删除覆盖并使用 Hermes/provider 默认值provider 的高级字段和 max_text_length 会保留在 raw YAML 中。', 'This writes tts.* and voice.*. API keys are still managed through .env. Empty optional voice IDs remove overrides and fall back to Hermes/provider defaults. Advanced provider fields and max_text_length stay in raw YAML.', '這裡寫入 tts.* 與 voice.*。API Key 仍透過 .env 管理;留空的可選聲音 ID 會刪除覆蓋並使用 Hermes/provider 預設值provider 的進階欄位和 max_text_length 會保留在 raw YAML 中。'),
hermesCompressionTitle: _('上下文压缩', 'Context compression', '上下文壓縮'),
hermesCompressionDesc: _('控制长对话何时触发压缩、压缩目标和保留范围,降低上下文过长导致的失败与费用浪费。', 'Control when long conversations are compressed, the target size, and protected message ranges to reduce failures and wasted cost from oversized context.', '控制長對話何時觸發壓縮、壓縮目標和保留範圍,降低上下文過長導致的失敗與費用浪費。'),
hermesCompressionStatusReady: _('结构化配置', 'structured settings', '結構化設定'),

View File

@@ -476,6 +476,33 @@ test('Hermes 配置页会暴露语音转写结构化配置字段', () => {
}
})
test('Hermes 配置页会暴露语音输出与录音结构化配置字段', () => {
for (const id of [
'hm-tts-voice-save',
'hm-tts-provider',
'hm-tts-edge-voice',
'hm-tts-openai-model',
'hm-tts-openai-voice',
'hm-tts-elevenlabs-voice-id',
'hm-tts-elevenlabs-model-id',
'hm-tts-xai-voice-id',
'hm-tts-xai-language',
'hm-tts-xai-sample-rate',
'hm-tts-xai-bit-rate',
'hm-tts-mistral-model',
'hm-tts-mistral-voice-id',
'hm-tts-piper-voice',
'hm-voice-record-key',
'hm-voice-max-recording-seconds',
'hm-voice-auto-tts',
'hm-voice-beep-enabled',
'hm-voice-silence-threshold',
'hm-voice-silence-duration',
]) {
assert.match(source, new RegExp(`id="${id}"`), `缺少 ${id}`)
}
})
test('Hermes 配置页会暴露 Kanban 调度稳定性结构化配置字段', () => {
for (const id of [
'hm-kanban-config-save',

View File

@@ -0,0 +1,183 @@
import test from 'node:test'
import assert from 'node:assert/strict'
import {
buildHermesTtsVoiceConfigValues,
mergeHermesTtsVoiceConfig,
} from '../scripts/dev-api.js'
test('Hermes TTS/Voice 配置读取会提供上游默认值', () => {
const values = buildHermesTtsVoiceConfigValues({})
assert.deepEqual(values, {
ttsProvider: 'edge',
ttsEdgeVoice: 'en-US-AriaNeural',
ttsOpenaiModel: 'gpt-4o-mini-tts',
ttsOpenaiVoice: 'alloy',
ttsElevenlabsVoiceId: 'pNInz6obpgDQGcFmaJgB',
ttsElevenlabsModelId: 'eleven_multilingual_v2',
ttsXaiVoiceId: 'eve',
ttsXaiLanguage: 'en',
ttsXaiSampleRate: 24000,
ttsXaiBitRate: 128000,
ttsMistralModel: 'voxtral-mini-tts-2603',
ttsMistralVoiceId: 'c69964a6-ab8b-4f8a-9465-ec0925096ec8',
ttsPiperVoice: 'en_US-lessac-medium',
voiceRecordKey: 'ctrl+b',
voiceMaxRecordingSeconds: 120,
voiceAutoTts: false,
voiceBeepEnabled: true,
voiceSilenceThreshold: 200,
voiceSilenceDuration: 3,
})
})
test('Hermes TTS/Voice 配置读取会回显 YAML 字段', () => {
const values = buildHermesTtsVoiceConfigValues({
tts: {
provider: 'openai',
edge: { voice: 'zh-CN-XiaoxiaoNeural' },
openai: { model: 'gpt-4o-mini-tts', voice: 'nova' },
elevenlabs: { voice_id: 'voice-123', model_id: 'eleven_turbo_v2_5' },
xai: {
voice_id: 'custom-eve',
language: 'zh',
sample_rate: 48000,
bit_rate: 192000,
},
mistral: { model: 'voxtral-mini-tts-2603', voice_id: 'mistral-voice' },
piper: { voice: 'zh_CN-huayan-medium' },
},
voice: {
record_key: 'ctrl+shift+v',
max_recording_seconds: 240,
auto_tts: true,
beep_enabled: false,
silence_threshold: 350,
silence_duration: 1.5,
},
})
assert.equal(values.ttsProvider, 'openai')
assert.equal(values.ttsEdgeVoice, 'zh-CN-XiaoxiaoNeural')
assert.equal(values.ttsOpenaiVoice, 'nova')
assert.equal(values.ttsElevenlabsVoiceId, 'voice-123')
assert.equal(values.ttsXaiLanguage, 'zh')
assert.equal(values.ttsXaiSampleRate, 48000)
assert.equal(values.ttsMistralVoiceId, 'mistral-voice')
assert.equal(values.ttsPiperVoice, 'zh_CN-huayan-medium')
assert.equal(values.voiceRecordKey, 'ctrl+shift+v')
assert.equal(values.voiceAutoTts, true)
assert.equal(values.voiceBeepEnabled, false)
assert.equal(values.voiceSilenceDuration, 1.5)
})
test('Hermes TTS/Voice 配置保存会保留未知字段并写入上游结构', () => {
const next = mergeHermesTtsVoiceConfig({
model: { provider: 'anthropic' },
tts: {
provider: 'edge',
custom_flag: 'keep-tts',
openai: { custom_flag: 'keep-openai' },
piper: { voices_dir: '/cache/piper' },
},
voice: {
custom_flag: 'keep-voice',
},
streaming: { enabled: true },
}, {
ttsProvider: 'openai',
ttsEdgeVoice: 'zh-CN-XiaoxiaoNeural',
ttsOpenaiModel: 'gpt-4o-mini-tts',
ttsOpenaiVoice: 'nova',
ttsElevenlabsVoiceId: 'voice-123',
ttsElevenlabsModelId: 'eleven_turbo_v2_5',
ttsXaiVoiceId: 'eve-pro',
ttsXaiLanguage: 'zh',
ttsXaiSampleRate: '48000',
ttsXaiBitRate: '192000',
ttsMistralModel: 'voxtral-mini-tts-2603',
ttsMistralVoiceId: 'mistral-voice',
ttsPiperVoice: 'zh_CN-huayan-medium',
voiceRecordKey: 'ctrl+shift+v',
voiceMaxRecordingSeconds: '240',
voiceAutoTts: true,
voiceBeepEnabled: false,
voiceSilenceThreshold: '350',
voiceSilenceDuration: '1.5',
})
assert.deepEqual(next.model, { provider: 'anthropic' })
assert.deepEqual(next.streaming, { enabled: true })
assert.equal(next.tts.provider, 'openai')
assert.equal(next.tts.edge.voice, 'zh-CN-XiaoxiaoNeural')
assert.equal(next.tts.openai.model, 'gpt-4o-mini-tts')
assert.equal(next.tts.openai.voice, 'nova')
assert.equal(next.tts.openai.custom_flag, 'keep-openai')
assert.equal(next.tts.elevenlabs.voice_id, 'voice-123')
assert.equal(next.tts.elevenlabs.model_id, 'eleven_turbo_v2_5')
assert.equal(next.tts.xai.sample_rate, 48000)
assert.equal(next.tts.xai.bit_rate, 192000)
assert.equal(next.tts.mistral.voice_id, 'mistral-voice')
assert.equal(next.tts.piper.voice, 'zh_CN-huayan-medium')
assert.equal(next.tts.piper.voices_dir, '/cache/piper')
assert.equal(next.tts.custom_flag, 'keep-tts')
assert.equal(next.voice.record_key, 'ctrl+shift+v')
assert.equal(next.voice.max_recording_seconds, 240)
assert.equal(next.voice.auto_tts, true)
assert.equal(next.voice.beep_enabled, false)
assert.equal(next.voice.silence_threshold, 350)
assert.equal(next.voice.silence_duration, 1.5)
assert.equal(next.voice.custom_flag, 'keep-voice')
})
test('Hermes TTS/Voice 配置保存空可选字段会删除对应覆盖', () => {
const next = mergeHermesTtsVoiceConfig({
tts: {
edge: { voice: 'custom-edge' },
elevenlabs: { voice_id: 'voice-123', model_id: 'model-123' },
piper: { voice: 'custom-piper', voices_dir: '/cache/piper' },
},
voice: {
record_key: 'ctrl+shift+v',
custom_flag: 'keep-voice',
},
}, {
ttsEdgeVoice: '',
ttsElevenlabsVoiceId: ' ',
ttsElevenlabsModelId: '',
ttsPiperVoice: '',
voiceRecordKey: '',
})
assert.equal(Object.hasOwn(next.tts.edge, 'voice'), false)
assert.equal(Object.hasOwn(next.tts.elevenlabs, 'voice_id'), false)
assert.equal(Object.hasOwn(next.tts.elevenlabs, 'model_id'), false)
assert.equal(Object.hasOwn(next.tts.piper, 'voice'), false)
assert.equal(next.tts.piper.voices_dir, '/cache/piper')
assert.equal(Object.hasOwn(next.voice, 'record_key'), false)
assert.equal(next.voice.custom_flag, 'keep-voice')
})
test('Hermes TTS/Voice 配置保存会拒绝非法枚举和越界值', () => {
assert.throws(
() => mergeHermesTtsVoiceConfig({}, { ttsProvider: 'bad' }),
/tts\.provider/,
)
assert.throws(
() => mergeHermesTtsVoiceConfig({}, { ttsOpenaiVoice: 'robot' }),
/tts\.openai\.voice/,
)
assert.throws(
() => mergeHermesTtsVoiceConfig({}, { ttsXaiSampleRate: '0' }),
/tts\.xai\.sample_rate/,
)
assert.throws(
() => mergeHermesTtsVoiceConfig({}, { voiceMaxRecordingSeconds: '0' }),
/voice\.max_recording_seconds/,
)
assert.throws(
() => mergeHermesTtsVoiceConfig({}, { voiceSilenceDuration: '-1' }),
/voice\.silence_duration/,
)
})