Generate Mandarin narration audio with word-level timestamps using Volcengine Doubao Speech 2.0.
npx skills add https://github.com/mooch10/mochiptos --skill doubao-tts-mooch10
Please help me install this Agent Skill. Skill: doubao-tts Source: https://github.com/mooch10/mochiptos/tree/main/frontend/.agents/doubao-tts Command: npx skills add https://github.com/mooch10/mochiptos --skill doubao-tts-mooch10
sentences[].words[] timing data saved as JSON, enabling precise caption generation without estimating text length. - Sample-First Workflow: Guides generating a short approval sample before committing to a full paid narration, with troubleshooting for auth, quota, and voice permission errors. - Use Case: You are producing a Chinese explainer video. Generate a 10-second sample voiceover for approval, then render the full narration and build subtitles from the returned word-level timestamps. ## Quick Start Ask the AI to generate a Mandarin narration of your script using the Doubao TTS provider with timestamps enabled and save the audio to your project's assets folder.High-intent search queries and answers about installing and using this skill.
Set DOUBAO_SPEECH_API_KEY in your .env file, then call the Doubao TTS provider with your text, a voice_id such as zh_female_vv_uranus_bigtts, and an output path. The async submit/query endpoints return the audio file plus timing metadata.
Enable the enable_timestamp parameter when calling the Doubao Speech 2.0 API. The query response contains sentences[].words[] timing data, saved as a JSON file next to the audio, which you use to build captions instead of estimating by character count.
This error means the wrong key type or auth header is being used. New-console API keys require the X-Api-Key header with X-Api-Resource-Id set to seed-tts-2.0, not the older X-Api-App-Id and X-Api-Access-Key headers.
Start with speech_rate 0, which is normal speed. A value of 100 doubles speed and -50 halves it. If the narration is too long, generate a short comparison sample at 25 or 50 before regenerating the full audio.
The speaker permission denied error occurs when the voice_id is incorrect or not authorized for the selected resource. Verify the voice type matches a Doubao Speech 2.0 voice and that your API key has access to the seed-tts-2.0 resource.
calesthioGenerate Mandarin narration audio with word-level timestamps using Volcengine Doubao Speech 2.0.
guanyangTranscribe, display, and animate captions in Remotion videos using @remotion/captions.
inference-shGenerate natural-sounding speech from text using multiple TTS models.
xingBaGanGenerate Mandarin and multilingual narration with Volcengine Doubao Speech 2.0 API, including subtitle-ready timing metadata.