elevenlabs

Create an ElevenLabs client with your API key and call client.text_to_speech.convert with your text, a voice_id, and a model like eleven_multilingual_v2. Save the returned audio stream to an MP3 file using the save helper or by writing chunks.

Use eleven_multilingual_v2 for stable, production-ready output in 29 languages. Choose eleven_flash_v2_5 or eleven_turbo_v2_5 when you need low latency and SSML break or phoneme tags. Use eleven_v3 only for maximum expressiveness, since it is alpha and less predictable.

With flash or turbo models, insert SSML break tags like <break time="1.5s" /> inline, up to 3 seconds each. Multilingual_v2 and v3 lack SSML support, so use blank lines for short pauses or insert silence with ffmpeg in post-processing.

Yes, use client.voices.ivc.create with one or more audio samples in binary mode for instant voice cloning. Professional Voice Cloning requires a Creator plan or higher, at least 30 minutes of clean audio, and a verification recording before training.

Generate one MP3 per scene plus a manifest.json containing each file's duration, then convert durations to frames by multiplying by your composition's fps. Wrap scenes in Series.Sequence components with those frame counts and add Audio components referencing the files.

Sound effects are capped at 22 seconds per generation and accept a text prompt plus optional duration and prompt_influence parameters. Music generation ranges from 10 seconds to 5 minutes and can be forced instrumental for background tracks.