Formerly KrillinAI, VideoTranslation and Dubbing Tool for 120+ Languages. Open-source AI workspace for creators, powered by Codex. Video downloader, video translation, create videos, images, audios, and edits with Agents in one place. For your Social Media like YouTube, TikTok, Instagram, etc.
TypeScript agent-skillscodex-clideskop-appdubbingimage-generationlocalizationskillsttsvideo-editingvideo-generationvideo-transcriptionvideo-translation
import type {
CreatorTtsProvider,
CreatorTtsVoice
} from '@opencreator/protocol';
import {
LoaderCircle,
Pause,
Play,
RefreshCw
} from 'lucide-react';
import { useCallback, useEffect, useMemo, useRef, useState } from 'react';
import { IssueList } from '../../features/issues/IssuePresenter.js';
import { usePageIssueState } from '../../features/issues/page-issue-state.js';
import { useAppLanguage } from '../../i18n/LanguageProvider.js';
import { useLocalizedCopy } from '../../i18n/useLocalizedCopy.js';
import type { CreatorServicesSettingsService } from '../../services/creator-services-service.js';
import { TtsVoiceSelect, type TtsVoiceSelectGroup } from './TtsVoiceSelect.js';
import { TtsVoiceLibrary, type VoiceLibraryOption } from './TtsVoiceLibrary.js';
import { TtsVoiceLibraryPopover } from './TtsVoiceLibraryPopover.js';
import { voiceScenarioLabel } from './voice-scenario-copy.js';
import './tts-voice-picker.css';
export function TtsVoicePicker(props: {
id: string;
provider: CreatorTtsProvider;
model: string;
catalogKey?: string;
value: string;
service: Pick<CreatorServicesSettingsService, 'getTtsVoices' | 'previewTtsVoice'> | null;
label?: string;
presentation?: 'select' | 'library';
disabled?: boolean;
allowCustomVoice?: boolean;
allowPreviewLanguage?: boolean;
onChange(voiceId: string, voice?: CreatorTtsVoice): void;
onVoiceResolved?(voice: CreatorTtsVoice): void;
}) {
const l = useLocalizedCopy();
const { language: appLanguage } = useAppLanguage();
const allowCustomVoice = props.allowCustomVoice ?? props.provider === 'volcengine';
const pageIssues = usePageIssueState('tts-voice');
const [voices, setVoices] = useState<CreatorTtsVoice[]>([]);
const [loading, setLoading] = useState(false);
const [previewing, setPreviewing] = useState(false);
const [playing, setPlaying] = useState(false);
const [previewVoiceId, setPreviewVoiceId] = useState('');
const [reloadToken, setReloadToken] = useState(0);
const [customId, setCustomId] = useState(props.value);
const [previewLanguageOverride, setPreviewLanguageOverride] = useState('');
const [sampleLanguage, setSampleLanguage] = useState('');
const audioRef = useRef<HTMLAudioElement>(null);
const previewUrlRef = useRef('');
const previewRequestRef = useRef(0);
const onVoiceResolvedRef = useRef(props.onVoiceResolved);
onVoiceResolvedRef.current = props.onVoiceResolved;
useEffect(() => {
let active = true;
if (props.service === null || props.provider === 'edge-tts') {
setVoices([]);
setLoading(false);
pageIssues.resolveOperation('tts.load-voices');
return () => {
active = false;
};
}
setLoading(true);
setVoices([]);
void props.service.getTtsVoices(props.provider, props.model)
.then(response => {
if (!active) return;
setVoices(response.voices);
pageIssues.resolveOperation('tts.load-voices');
})
.catch(cause => {
if (active) {
setVoices([]);
pageIssues.captureOperationFailure(
'tts.load-voices',
cause,
l('无法加载音色列表,请检查服务配置和网络后重试。', 'Could not load voices. Check the service settings and network, then retry.'),
{ retryable: true }
);
}
})
.finally(() => {
if (active) setLoading(false);
});
return () => {
active = false;
};
}, [
l,
pageIssues.captureOperationFailure,
pageIssues.resolveOperation,
props.model,
props.catalogKey,
props.provider,
props.service,
reloadToken
]);
useEffect(() => {
const audio = audioRef.current;
return () => {
previewRequestRef.current += 1;
if (audio?.getAttribute('src')) audio.pause();
if (previewUrlRef.current) URL.revokeObjectURL(previewUrlRef.current);
};
}, []);
useEffect(() => {
setCustomId(props.value);
}, [props.value]);
const options = useMemo(() => {
if (!props.value || voices.some(voice => voice.id === props.value)) return voices;
return [{
id: props.value,
name: props.value,
provider: props.provider,
kind: 'custom' as const
}, ...voices];
}, [props.provider, props.value, voices]);
const previewLanguages = useMemo(() => [...new Set(voices.flatMap(voice => voice.previewLanguages ?? []))]
.sort((a, b) => Number(b === appLanguage.split('-')[0]) - Number(a === appLanguage.split('-')[0])
|| Number(b === 'en') - Number(a === 'en')
|| (voiceLanguageLabel(a, appLanguage, l) ?? a).localeCompare(voiceLanguageLabel(b, appLanguage, l) ?? b, appLanguage)), [voices, appLanguage, l]);
const previewLanguage = props.allowPreviewLanguage
? availablePreviewLanguage(previewLanguages, previewLanguageOverride, appLanguage)
: appLanguage;
const visibleOptions = useMemo(() => props.allowPreviewLanguage && previewLanguage
? options.filter(voice => voice.previewLanguages?.includes(previewLanguage))
: options, [options, previewLanguage, props.allowPreviewLanguage]);
const selectGroups = useMemo<TtsVoiceSelectGroup[]>(() => groupVoices(visibleOptions).map(group => ({
label: group.label ? voiceScenarioLabel(group.label, l) : '',
options: group.voices.map(voice => ({
value: voice.id,
name: voiceName(voice, l),
detail: voiceDetail(voice, l, appLanguage, group.label)
}))
})), [appLanguage, l, visibleOptions]);
const selected = options.find(voice => voice.id === props.value);
const selectedHasPreview = !props.allowPreviewLanguage || !!previewLanguage && !!selected?.previewLanguages?.includes(previewLanguage);
const libraryOptions = useMemo<VoiceLibraryOption[]>(() => visibleOptions.map(voice => ({
id: voice.id,
name: voice.name,
language: voiceLanguageLabel(voice.language, appLanguage, l) ?? '',
gender: voiceGenderLabel(voice.gender, l),
traits: localizedVoiceInfo(voice.traits, appLanguage) ?? [],
useCases: localizedVoiceInfo(voice.useCases, appLanguage) ?? [],
description: localizedVoiceInfo(voice.description, appLanguage),
scenario: voice.scenario ? voiceScenarioLabel(voice.scenario, l) : '',
kind: voiceKindLabel(voice.kind, l),
custom: voice.kind === 'custom',
recommended: voice.recommended === true,
previewAvailable: !props.allowPreviewLanguage || !!voice.previewLanguages?.length,
search: [voice.id, voice.name, voice.language, voice.gender, voice.scenario,
...Object.values(voice.traits ?? {}).flat(), ...Object.values(voice.useCases ?? {}).flat(),
...Object.values(voice.description ?? {}),
voiceLanguageLabel(voice.language, appLanguage, l), voiceGenderLabel(voice.gender, l),
voice.scenario ? voiceScenarioLabel(voice.scenario, l) : '', voiceKindLabel(voice.kind, l)].join(' ')
})), [appLanguage, l, visibleOptions, props.allowPreviewLanguage]);
useEffect(() => {
if (selected) onVoiceResolvedRef.current?.(selected);
}, [selected]);
const stopPreview = useCallback(() => {
previewRequestRef.current += 1;
if (audioRef.current?.getAttribute('src')) {
audioRef.current.pause();
audioRef.current.removeAttribute('src');
}
if (previewUrlRef.current) URL.revokeObjectURL(previewUrlRef.current);
previewUrlRef.current = '';
setPlaying(false);
setPreviewing(false);
setPreviewVoiceId('');
}, []);
useEffect(() => {
stopPreview();
setSampleLanguage('');
}, [props.provider, props.model, props.value, props.disabled, props.service, appLanguage, previewLanguage, stopPreview]);
function selectVoice(voiceId: string) {
stopPreview();
const voice = options.find(candidate => candidate.id === voiceId);
setCustomId(voiceId);
if (voiceId !== props.value) props.onChange(voiceId, voice);
}
function commitCustomVoice(raw: string) {
const voiceId = raw.trim();
setCustomId(voiceId);
if (voiceId === props.value) return;
stopPreview();
const voice = options.find(candidate => candidate.id === voiceId);
props.onChange(voiceId, voice ?? {
id: voiceId,
name: voiceId,
provider: props.provider,
kind: 'custom'
});
}
async function preview(voiceId = props.value) {
if (previewVoiceId === voiceId && (playing || previewing)) {
stopPreview();
return;
}
if (
props.service === null
|| props.provider === 'edge-tts'
|| !voiceId
|| props.disabled
|| loading
) return;
const language = previewLanguage;
if (props.allowPreviewLanguage && !options.find(voice => voice.id === voiceId)?.previewLanguages?.includes(language)) return;
if (!language) return;
stopPreview();
setSampleLanguage('');
const request = ++previewRequestRef.current;
setPreviewVoiceId(voiceId);
setPreviewing(true);
try {
const response = await props.service.previewTtsVoice({
provider: props.provider,
model: props.model,
language,
voiceId
});
if (!response.ok) throw new Error(`Voice preview request failed: ${response.status}`);
const blob = await response.blob();
if (request !== previewRequestRef.current) return;
const url = URL.createObjectURL(blob);
if (previewUrlRef.current) URL.revokeObjectURL(previewUrlRef.current);
previewUrlRef.current = url;
const audio = audioRef.current;
if (!audio) return;
audio.src = url;
await audio.play();
if (request !== previewRequestRef.current) return;
setSampleLanguage(response.headers.get('Content-Language') || '');
setPlaying(true);
pageIssues.resolveOperation('tts.preview-voice');
} catch (cause) {
if (request !== previewRequestRef.current) return;
pageIssues.captureOperationFailure(
'tts.preview-voice',
cause,
l('音色试听失败,请稍后重试。', 'Voice preview failed. Please try again later.'),
{ retryable: true }
);
} finally {
if (request === previewRequestRef.current) setPreviewing(false);
}
}
const previewLanguageControl = props.allowPreviewLanguage ? <div className="tts-voice-preview-language">
<span id={`${props.id}-preview-language-label`}>{l('试听语言', 'Preview language')}</span>
<TtsVoiceSelect id={`${props.id}-preview-language`} labelId={`${props.id}-preview-language-label`} value={previewLanguage}
groups={[{ label: '', options: previewLanguages.map(language => {
const name = voiceLanguageLabel(language, appLanguage, l) ?? language;
const english = voiceLanguageLabel(language, 'en', l);
return { value: language, name, detail: english !== name ? english : language.toUpperCase() };
}) }]}
placeholder={loading ? l('正在加载语言', 'Loading languages') : l('暂无可试听语言', 'No preview languages available')}
searchLabel={l('搜索语言', 'Search languages')}
emptyLabel={l('未找到匹配语言', 'No matching languages')}
disabled={props.disabled || loading || !previewLanguages.length || props.service === null || props.provider === 'edge-tts'}
onChange={language => { stopPreview(); setSampleLanguage(''); setPreviewLanguageOverride(language); }} />
<small role="status">{sampleLanguage
? sampleLanguage.toLowerCase().split('-')[0] !== previewLanguage.toLowerCase().split('-')[0]
? l(`所选语言暂无样音,当前试听:${voiceLanguageLabel(sampleLanguage, appLanguage, l)}`, `No sample for the selected language. Playing: ${voiceLanguageLabel(sampleLanguage, appLanguage, l)}`)
: l(`当前试听:${voiceLanguageLabel(sampleLanguage, appLanguage, l)}`, `Playing: ${voiceLanguageLabel(sampleLanguage, appLanguage, l)}`)
: l('免费试听,选好语言后再选择音色。', 'Free previews. Choose a language, then a voice.')}</small>
</div> : null;
const voiceDetails = <>
{allowCustomVoice ? (
<label className="tts-voice-picker-custom" htmlFor={`${props.id}-custom`}>
<span>{l('克隆 / 自定义 Speaker ID', 'Clone / custom speaker ID')}</span>
<input
id={`${props.id}-custom`}
type="text"
value={customId}
placeholder={l('例如 S_xxxxxxxx 或官方音色 ID', 'e.g. S_xxxxxxxx or an official voice ID')}
spellCheck={false}
autoComplete="off"
disabled={props.disabled || props.provider === 'edge-tts'}
onChange={event => setCustomId(event.target.value)}
onBlur={() => commitCustomVoice(customId)}
onKeyDown={event => {
if (event.key === 'Enter') {
event.preventDefault();
commitCustomVoice(customId);
}
}}
/>
</label>
) : null}
{!selected && allowCustomVoice && !loading ? (
<small className="tts-voice-picker-meta">
{l(
'声音复刻请填写控制台中的 Speaker ID(S_ 开头)。',
'For cloned voices, enter the console Speaker ID starting with S_.'
)}
</small>
) : null}
<IssueList
issues={pageIssues.issues}
actions={{
retryOperations: {
'tts.load-voices': () => setReloadToken(current => current + 1),
'tts.preview-voice': () => preview(previewVoiceId || props.value)
}
}}
onDismiss={pageIssues.dismissIssue}
compact
/>
</>;
return (
<div className={`tts-voice-picker${props.presentation === 'library' ? ' is-library' : ''}`}>
{previewLanguageControl}
{props.presentation === 'library' ? <TtsVoiceLibraryPopover
label={props.label ?? l('默认音色', 'Default voice')}
name={loading ? l('正在加载音色', 'Loading voices') : selected?.name ?? props.value}
detail={selected && !selectedHasPreview && previewLanguage
? l('此语言暂无该音色样音,可重新选择音色', 'No sample in this language. Choose another voice to preview.')
: selected ? voiceDetail(selected, l, appLanguage, '') : undefined}
disabled={props.disabled || props.provider === 'edge-tts'}
resetKey={`${props.provider}:${props.model}:${appLanguage}:${previewLanguage}`}
onClose={stopPreview}
>{close => <>
<TtsVoiceLibrary
key={`${props.provider}:${props.model}:${appLanguage}:${previewLanguage}`}
id={props.id}
label={props.label ?? l('默认音色', 'Default voice')}
value={props.value}
options={libraryOptions}
loading={loading}
disabled={props.disabled || props.provider === 'edge-tts'}
canRefresh={props.service !== null && props.provider !== 'edge-tts'}
canPreview={props.service !== null && props.provider !== 'edge-tts'}
previewVoiceId={previewVoiceId}
previewing={previewing}
playing={playing}
onRefresh={() => { stopPreview(); setReloadToken(current => current + 1); }}
onPreview={voiceId => void preview(voiceId)}
onStopPreview={stopPreview}
onChange={(voiceId, commit) => { selectVoice(voiceId); if (commit) close(); }}
onClose={close}
/>
{voiceDetails}
</>}</TtsVoiceLibraryPopover> : <>
<div className="tts-voice-picker-field">
<span id={`${props.id}-label`}>{props.label ?? l('默认音色', 'Default voice')}</span>
<TtsVoiceSelect
id={props.id}
labelId={`${props.id}-label`}
value={props.value}
fallbackName={selected ? voiceName(selected, l) : props.value || undefined}
groups={selectGroups}
placeholder={loading
? l('正在加载音色', 'Loading voices')
: allowCustomVoice
? l('请填写下方 Speaker ID', 'Enter a speaker ID below')
: l('暂无可用音色', 'No voices available')}
disabled={props.disabled || loading || props.provider === 'edge-tts'}
onChange={selectVoice}
/>
</div>
<div className="tts-voice-picker-actions">
<button
type="button"
disabled={props.disabled || loading || props.service === null || props.provider === 'edge-tts'}
aria-label={l('刷新音色列表', 'Refresh voices')}
title={l('刷新音色列表', 'Refresh voices')}
onClick={() => { stopPreview(); setReloadToken(current => current + 1); }}
>
{loading
? <LoaderCircle className="tts-voice-picker-spinner" size={16} aria-hidden="true" />
: <RefreshCw size={16} aria-hidden="true" />}
</button>
<button
type="button"
disabled={props.disabled || loading || !selected || props.service === null || props.provider === 'edge-tts' || previewing || !selectedHasPreview}
aria-label={playing ? l('暂停音色试听', 'Pause voice preview') : l('试听当前音色', 'Preview selected voice')}
title={playing ? l('暂停', 'Pause') : l('试听', 'Preview')}
onClick={() => void preview()}
>
{previewing
? <LoaderCircle className="tts-voice-picker-spinner" size={16} aria-hidden="true" />
: playing
? <Pause size={16} aria-hidden="true" />
: <Play size={16} aria-hidden="true" />}
</button>
</div>
</>}
<audio ref={audioRef} hidden onEnded={stopPreview} />
{props.presentation !== 'library' ? voiceDetails : null}
</div>
);
}
function availablePreviewLanguage(languages: string[], preferred: string, appLanguage: string): string {
return languages.find(language => language === preferred.split('-')[0])
?? languages.find(language => language === appLanguage.split('-')[0])
?? languages.find(language => language === 'en')
?? languages[0] ?? '';
}
function voiceName(voice: CreatorTtsVoice, l: (zh: string, en: string) => string): string {
const name = voice.name === voice.id ? voice.name : `${voice.name} (${voice.id})`;
return voice.recommended ? `${name} · ${l('推荐', 'Recommended')}` : name;
}
function voiceDetail(
voice: CreatorTtsVoice,
l: (zh: string, en: string) => string,
appLanguage: string,
groupLabel: string
): string | undefined {
// The group header already shows the scenario, so only repeat it for ungrouped voices.
const details = [
voiceGenderLabel(voice.gender, l),
...(localizedVoiceInfo(voice.traits, appLanguage) ?? []).slice(0, 2),
voice.scenario && voice.scenario !== groupLabel ? voiceScenarioLabel(voice.scenario, l) : undefined,
voiceLanguageLabel(voice.language, appLanguage, l),
voiceKindLabel(voice.kind, l)
].filter(Boolean);
return details.length > 0 ? details.join(' · ') : undefined;
}
function localizedVoiceInfo<T>(values: Record<string, T> | undefined, language: string): T | undefined {
return values?.[language] ?? values?.[language.split('-')[0]!] ?? values?.en;
}
function voiceLanguageLabel(
code: string | undefined,
appLanguage: string,
l: (zh: string, en: string) => string
): string | undefined {
if (!code) return undefined;
if (code === 'multi') return l('多语种', 'Multilingual');
const dialect = chineseDialects[code];
if (dialect !== undefined) return l(dialect[0], dialect[1]);
// Catalogs mix BCP 47 codes with free-form labels; anything Intl rejects is shown as-is.
try {
const names = new Intl.DisplayNames([appLanguage], { type: 'language', fallback: 'none' });
return names.of(code) ?? code;
} catch {
return code;
}
}
const chineseDialects: Readonly<Record<string, readonly [string, string]>> = {
'zh-yue': ['粤语', 'Cantonese'],
'zh-minnan': ['闽南语', 'Hokkien'],
'zh-sichuan': ['四川话', 'Sichuanese'],
'zh-dongbei': ['东北话', 'Northeastern Mandarin'],
'zh-beijing': ['北京话', 'Beijing dialect'],
'zh-tianjin': ['天津话', 'Tianjin dialect'],
'zh-shanghai': ['上海话', 'Shanghainese'],
'zh-nanjing': ['南京话', 'Nanjing dialect'],
'zh-shaanxi': ['陕西话', 'Shaanxi dialect'],
'zh-xian': ['西安话', "Xi'an dialect"],
'zh-hunan': ['湖南话', 'Hunanese'],
'zh-guangxi': ['广西话', 'Guangxi dialect'],
'zh-zhengzhou': ['郑州话', 'Zhengzhou dialect']
};
function voiceKindLabel(
kind: CreatorTtsVoice['kind'],
l: (zh: string, en: string) => string
): string {
if (kind === 'custom') return l('自定义音色', 'Custom');
if (kind === 'designed') return l('设计音色', 'Designed');
return '';
}
function voiceGenderLabel(gender: string | undefined, l: (zh: string, en: string) => string): string {
if (!gender) return '';
if (/^(?:male|男|男性|男声)$/i.test(gender)) return l('男声', 'Male');
if (/^(?:female|女|女性|女声)$/i.test(gender)) return l('女声', 'Female');
if (/^(?:neutral|中性)$/i.test(gender)) return l('中性', 'Neutral');
return gender;
}
function groupVoices(voices: CreatorTtsVoice[]): Array<{
label: string;
voices: CreatorTtsVoice[];
}> {
const groups: Array<{ label: string; voices: CreatorTtsVoice[] }> = [];
const index = new Map<string, number>();
for (const voice of voices) {
const label = voice.kind === 'custom' ? '' : (voice.scenario ?? '');
const existing = index.get(label);
if (existing === undefined) {
index.set(label, groups.length);
groups.push({ label, voices: [voice] });
continue;
}
groups[existing]!.voices.push(voice);
}
return groups;
}