voice

Voice & Sound Effects Generator

Generate voiceovers (TTS) and sound effects. For TTS, choose a concrete provider and voice before calling submit_voice.

When to Use

• Generate voiceover/narration from text
• Create text-to-speech audio for videos
• Add, replace, or redo narration/voiceover for an existing video, timeline, screen recording, slide animation, product demo, B-roll edit, MG explainer, or other visual sequence
• Keep existing narration/voiceover aligned after trimming, speeding up, slowing down, moving, reordering, or replacing the visuals it describes
• Offer and audition TTS voice choices when the user has not picked a concrete voice
• Generate custom sound effects from text descriptions only after checking the Sound Effects library first

TTS (Text-to-Speech)

If the current request has an existing visual target and the user wants narration, voiceover, dubbing, or replacement speech for that target, read references/video-sync.md before drafting new narration, using existing narration text to generate TTS, or placing audio. Do this even when the user did not explicitly say "sync" or "match the visuals"; the existence of a visual target means narration timing and meaning may need to follow on-screen content. Use the normal standalone TTS path only when there is no visual target or the user just wants an audio asset from text.

Also read references/video-sync.md when the timeline already has narration/voiceover and the user asks to change the visuals while keeping that voiceover aligned. This is a sync maintenance task even if no new TTS is needed.

Use submit_voice to create a TTS audio asset. The current MCP tool contract is:

• provider is required. Configured choices may be doubao, elevenlabs, minimax, inworld, fishaudio, speechify, openai, gemini, mistral, or cartesia. All providers are opt-in; use only providers shown as configured in the capabilities prompt.
• voiceId is required, concrete, and provider-specific. The only exception is deliberate MiniMax timbreWeights mixing, where voiceId must be empty. Do not mix catalogs.
• The curated catalog in references/voices.md covers only Doubao, ElevenLabs, and MiniMax. Other providers have no bundled preset or sample catalog in OpenChatCut. Require a concrete voice ID from the user or their provider account; never invent a preset or /voice-samples/... URL.
• AI SDK-backed fields are provider-specific: OpenAI supports modelId, speed, outputFormat, and instructions; Gemini supports modelId, outputFormat, and instructions; Mistral supports modelId and outputFormat; Cartesia supports modelId, speed, languageCode, and outputFormat. Omit unsupported or unrequested fields.
• Inworld, Fish Audio, and Speechify accept only voiceId plus optional modelId. Do not pass expressive, speed, language, or output controls to these providers.
• submit_voice creates an audio asset only. Timeline placement, replacement, trimming, and alignment happen later with timeline tools.
• For long narration, multiple submit_voice calls can be useful: split at natural pauses, sentence groups, or script beat boundaries when the workflow benefits from separately timed or placed voice clips.
• Doubao supports speedRatio, loudnessRatio, pitch, emotion, emotionScale, performancePrompt, and explicitDialect, but not every voice supports every expressive control. Check references/voices.md before using them.
• ElevenLabs retains its official voice settings, language, seed, output, normalization, pronunciation-dictionary, continuity, logging, and latency controls. MiniMax retains its dedicated controls documented in references/minimax-tts.md.

Doubao control support for current curated voices:

• vivi, xiaohe, yunzhou, xiaotian, naiqimengwa, yingtaowanzi, wenroumama, zhixingnv, dayi, jitangnv, liuchang, ruyayichen, morgan, qingcang, huiben, popo, yuanboxiaoshu, baqiqingshu, and tangseng support explicit emotion / emotionScale, performancePrompt, and ASMR-style prompt directions.
• shuanglangshaonian supports performancePrompt and COT/QA-style instruction following, but does not support explicit emotion / emotionScale or ASMR-style control.
• explicitDialect is only supported by vivi and can be dongbei, shaanxi, or sichuan.

ElevenLabs control support for current curated voices:

• amelia, brittney, hope, jessica, arabella, jane, maria, mark, frederick, peter, james, jon, sully, david, and alex all support the same request-level controls; model-specific support is still validated by ElevenLabs.
• These controls are not per-voice guarantees of a specific acting style. Use the preset tags/samples to pick a naturally suitable voice, then use the controls for moderate delivery changes.
• For ElevenLabs eleven_v3, inline audio tags are available when the user asks for expressive delivery such as emotion, tone, nonverbal cues, accent hints, or local pacing. Official examples fit these useful TTS categories: emotion/tone tags such as [happy], [sad], [angry], [excited], [curious], [sarcastic], [crying], [annoyed], [appalled], [thoughtful], [surprised], and [mischievously]; vocal delivery and nonverbal cue tags such as [whispers], [laughs], [sighs], [exhales], [inhales deeply], [clears throat], [snorts], [swallows], [wheezing], and [coughs]; pacing/pause/local speed tags such as [slowly], [pause], [short pause], [long pause], [rushed], and [drawn out]; and accent/special-performance tags such as [strong X accent], for example [strong French accent], plus [sings], [singing], [woo], and [pirate voice]. Official examples are non-exhaustive; similar auditory tags can be tried when the user explicitly asks for that delivery and the tag describes how the voice should sound, not a visual action. Write tags directly in text, close to the short phrase they should affect. Treat tags as local guidance, not paragraph-wide controls.
• For pauses and pacing in eleven_v3, use punctuation, text structure, shorter generated segments, or local audio tags such as [short pause] and [slowly] when needed.

// English / multilingual via ElevenLabs
submit_voice({
  provider: "elevenlabs",
  text: "Hello world",
  voiceId: "peter",
});

// Chinese via Doubao
submit_voice({
  provider: "doubao",
  text: "你好世界",
  voiceId: "liuchang",
});

// With speed adjustment (Doubao only)
submit_voice({
  provider: "doubao",
  text: "这是一段稍快的中文旁白。",
  voiceId: "liuchang",
  speedRatio: 1.5,
});

// With expressive Doubao controls
submit_voice({
  provider: "doubao",
  text: "这次事故提醒我们，安全永远不能侥幸。",
  voiceId: "liuchang",
  emotion: "sad",
  emotionScale: 3,
  performancePrompt: "痛心但克制，语速稍慢，像新闻专题旁白",
  pitch: -1,
  speedRatio: 0.92,
});

// With ElevenLabs delivery controls
submit_voice({
  provider: "elevenlabs",
  text: "The launch changed how teams plan their daily work.",
  voiceId: "peter",
  speed: 0.95,
  stability: 0.4,
  similarityBoost: 0.8,
  outputFormat: "wav_44100",
});

// MiniMax TTS (when configured) — see references/minimax-tts.md
submit_voice({
  provider: "minimax",
  text: "欢迎使用视频编辑助手。",
  voiceId: "female-yujie",
  speed: 1,
  name: "VO · welcome",
});

// Cartesia shape after the user confirms the exact account voice ID.
// confirmedCartesiaVoiceId represents that supplied value, not a preset.
submit_voice({
  provider: "cartesia",
  text: "A concise product introduction.",
  voiceId: confirmedCartesiaVoiceId,
  modelId: "sonic-3",
  speed: 1,
  languageCode: "en",
  outputFormat: "mp3",
});

Voice Audition Before Generation

When the user needs TTS and has not already chosen a concrete voice, first separate providers with curated OpenChatCut choices from providers that require an account-specific voice ID.

For Doubao, ElevenLabs, or MiniMax, read references/voices.md before recommending, rendering, or submitting an option. Use it as the only source for curated preset IDs, provider choice, display labels, tags, and bundled sample URLs. Do not create voice options from memory, translated names, or broad user descriptions.

For Inworld, Fish Audio, Speechify, OpenAI, Gemini, Mistral, or Cartesia, do not offer an invented audition list or sample URL. Ask the user for the concrete voice ID from that configured provider. A broad description such as "warm female" is not a valid voiceId.

First determine two separate languages:

• User conversation language: the language the user used to talk to you. Use this for the surrounding reply, form-visual label, visual-option name, and summary.
• Target narration language: the language of the text being synthesized. Use this only to choose provider and voice catalog.

The audition widget's submit button is fixed to the default label in this build (submitLabel is accepted but not rendered); keep the question label and option labels in the user conversation language, not the target narration language. For example: English users see submit_label="Submit", Chinese users see submit_label="提交", and Spanish users see submit_label="Enviar".

"help me generate ... voice over in Chinese" is an English conversation asking for Chinese narration, so the audition widget copy stays in English while the voice candidates come from Doubao.

For a curated provider:

1. Filter references/voices.md by target narration language / provider and explicit requirements such as gender, age range, tone, and use case.
2. If no preset matches all explicit requirements, say there is no exact match and offer the closest supported presets with a clear caveat.
3. Pick 2-4 matching curated presets.
4. Load widget-forms, then call ask_followup_questions with voice options and real bundled audio samples.
5. Wait for the user to choose.
6. Call submit_voice with the selected preset ID as voiceId.

For a provider without a curated OpenChatCut catalog, ask for a free-text, concrete provider voice ID instead. Do not add media or synthesize a /voice-samples/... path. Wait for the user to supply/confirm the exact ID before calling submit_voice.

For each curated audition option, keep value, display label, media, and summary tied to the same preset row from references/voices.md. Use only the sample URLs recorded there. Keep value as the preset ID and media as its matching sample URL. Write name and summary in the user's conversation language. The target narration language only decides the provider/voice catalog. After submission, map the display name back to the preset ID from the same candidate list.

English request for Chinese narration:

<widget submit_label="Submit">
  <form-visual
    id="voiceId"
    label="For Chinese voiceover, I recommend a few voices to try:"
    required="true"
  >
    <visual-option
      value="vivi"
      name="Vivi"
      media="/voice-samples/doubao-vivi.mp3"
      aspect-ratio="16:5"
      summary="Female / young / friendly, general"
    />
    <visual-option
      value="xiaohe"
      name="Xiaohe"
      media="/voice-samples/doubao-xiaohe.mp3"
      aspect-ratio="16:5"
      summary="Female / young / soft, clear"
    />
    <visual-option
      value="yunzhou"
      name="Yunzhou"
      media="/voice-samples/doubao-yunzhou.mp3"
      aspect-ratio="16:5"
      summary="Male / young / neutral, business"
    />
  </form-visual>
</widget>

Chinese request for Chinese narration:

<widget submit_label="提交">
  <form-visual
    id="voiceId"
    label="我推荐这几个中文旁白音色，先试听一下："
    required="true"
  >
    <visual-option
      value="morgan"
      name="Morgan"
      media="/voice-samples/doubao-morgan.mp3"
      aspect-ratio="16:5"
      summary="男 / 中年 / 低沉知识解说"
    />
    <visual-option
      value="zhixingnv"
      name="知性女声"
      media="/voice-samples/doubao-zhixingnv.mp3"
      aspect-ratio="16:5"
      summary="女 / 中年 / 冷静知识讲解"
    />
    <visual-option
      value="vivi"
      name="Vivi"
      media="/voice-samples/doubao-vivi.mp3"
      aspect-ratio="16:5"
      summary="女 / 年轻 / 亲切通用口播"
    />
  </form-visual>
</widget>

Sound Effects

For ordinary editing sound effects (SFX), do not generate first. Use the built-in Sound Effects library before generating:

1. Call browse_library with category:"sound-effects" and a query such as "whoosh", "camera shutter", "notification", "censor beep", or "record scratch".
2. Inspect the returned library:sound:<id>.
3. Place it with edit_item, using fromFrame as the sound's anchor/editorial moment frame:

browse_library({
  category: "sound-effects",
  query: "short whoosh transition",
});

edit_item({
  adds: [
    {
      type: "audio",
      assetId: "library:sound:whoosh-short",
      fromFrame: 120,
      trackId: "A1",
    },
  ],
});

Only generate sound effects from text descriptions with submit_sound when:

• The user explicitly asks for a generated/original/custom sound.
• The requested sound is too specific for the existing Sound Effects library.
• browse_library({ category:"sound-effects", query }) returns no suitable match.

// Custom/generated sound effect after the library has no suitable match
submit_sound({ prompt: "A dog barking in the distance" });

// With custom duration (0.5-22 seconds)
submit_sound({
  prompt: "Thunder and heavy rain",
  durationSeconds: 15,
});

// High prompt adherence
submit_sound({
  prompt: "Sci-fi laser gun firing",
  promptInfluence: 0.8,
});

Tips for better results:

• Be specific: "A dog barking loudly" vs just "dog"
• Include context: "Footsteps on wooden floor in an empty room"
• Specify style: "Cinematic whoosh" or "8-bit game sound"

Parameters

TTS

Sound Effects

Voices

Use the submit_voice voiceId guide and references/voices.md for the current curated preset list, display labels, tags, and sample URLs.

Voice IDs are provider-specific — do NOT mix them

The curated catalog contains separate Doubao, ElevenLabs, and MiniMax IDs. vivi / dayi are only Doubao; mark / amelia / james are only ElevenLabs; female-yujie is only MiniMax. Inworld, Fish Audio, Speechify, OpenAI, Gemini, Mistral, and Cartesia require a concrete provider-specific ID confirmed by the user and have no bundled OpenChatCut samples.

Provider choice:

• Honor an explicit configured provider first.
• For Chinese narration, prefer a matching curated Doubao voice (or MiniMax when configured/requested). Use another provider only after the user chooses it and confirms its voice ID.
• For English / multilingual narration, prefer a curated ElevenLabs voice. Use another provider only after the user chooses it and confirms its voice ID.
• Never offer a provider that is not shown as configured in capabilities.

Hard rules — what you must NOT do

1. Never use a voice ID from a different provider.
2. Never submit TTS while the voice is only described broadly; require a concrete provider-specific ID confirmed by the user.
3. Never recommend or render a curated TTS option before checking references/voices.md.
4. Never invent presets or sample URLs for Inworld, Fish Audio, Speechify, OpenAI, Gemini, Mistral, or Cartesia.
5. Never pass provider-specific fields to a provider that does not support them.
6. Never claim stable age, regional accent, pronunciation dictionary, or exact duration controls unless the selected provider exposes them.
7. Never replace original recorded speech with TTS unless the user asks.