Turn text into high-quality speech through a single API that gives you three voice providers to choose from:
Learn more about Text to Speech91 characters
Your audio shows up here
Pick a provider and voice, then hit Speak.
Turn text into high-quality speech through a single API that gives you three voice providers to choose from: ElevenLabs (the default, with the widest language and gender coverage), MiniMax (many languages with descriptive voice names), and OpenAI (six ready-made, language-agnostic voices). Pick a voice by language + gender from the shared catalog (the same one the Dubbing service uses), or pass an explicit voice_id for precise control. Optional voice_settings (stability, similarity_boost, and more) and an hd_quality flag let you fine-tune the result. Each request returns a URL to the generated audio file.
ElevenLabs (named voices plus a language/gender catalog and voice cloning), MiniMax (descriptive voice names per language), and OpenAI (six fixed voices: alloy, echo, fable, onyx, nova, shimmer). Switch between them with the provider field.
Pass language + gender and we pick the best matching voice from the provider's catalog. ElevenLabs and MiniMax cover 13+ languages; OpenAI's voices are fixed and language-agnostic.
Pass clone_from_audio (a URL) or clone_from_video to clone a voice from a sample for the request. Available on ElevenLabs and MiniMax — check the supports_voice_cloning flag in the catalog.
Tune stability (0-1), similarity_boost (0-1), speed, pitch (-12 to +12 semitones), volume (0.1-10), and emotion. Supported settings vary by provider — see supports_voice_settings in the catalog response.
Set hd_quality:true to use a higher-fidelity model where the provider supports it. Same per-second pricing.
Video narration
Generate voiceovers for tutorials, product demos, explainers without hiring a voice actor.
Accessibility
Convert articles, e-books, or documentation to audio for visually impaired users.
IVR / phone systems
Generate dynamic voice prompts in 13+ languages.
E-learning
Audio lessons and lecture narrations across languages for online courses.
input
Text + provider + (language + gender) OR voice_id; optional voice_settings, speed, hd_quality, clone_from_audio/video
output
A URL to the generated audio plus its duration. You're billed on usage.units, which equals the audio length in seconds.
For each request, the voice is chosen in this order:
voice_id, that voice is used directly (it takes precedence over everything else). It is the only voice field, and every provider accepts it — an ElevenLabs id or friendly name, a MiniMax descriptive name, or an OpenAI voice name.language + gender selects a voice from the catalog (ElevenLabs and MiniMax).voice_id it speaks with alloy.Pass clone_from_audio (a URL) to clone a voice from an audio sample just for that request. Set keep_source_file: true (with an optional auto_delete_after_hours) to keep the uploaded sample around longer. Per-call cloning isn't stored — for a reusable custom voice, use the Voice Clone service (voice-clone-tts).
stability (0-1): how consistent the voice stays across the output.similarity_boost (0-1): how closely the output matches the target voice.speed: speed multiplier (doesn't change billing).pitch (-12 to 12): semitones up or down.volume (0.1-10): output volume multiplier.emotion: an emotional tone hint (provider-specific).Settings a provider doesn't support are ignored.
textrequiredstringText to convert to speech (max ~5000 characters)
providerstring · default: elevenlabsSynthesis provider
enum: elevenlabs, minimax, openai
languagestringLanguage name from the provider's catalog (e.g. 'english', 'turkish'). Ignored when voice_id is set OR provider is openai (language-agnostic).
genderstringVoice gender. Ignored when voice_id is set OR provider is openai.
enum: male, female
voice_idstringThe voice to speak with — the one field every provider accepts. ElevenLabs: a voice ID (e.g. '21m00Tcm4TlvDq8ikWAM') or one of the friendly names (rachel, adam, antoni, arnold, bella, domi, elli, josh, nicole, sam). MiniMax: a descriptive voice name (e.g. 'English_CalmWoman'). OpenAI: a voice name (alloy, echo, fable, onyx, nova, shimmer). Takes precedence over language/gender. Call core-tts-voices for the full per-provider catalog.
speednumber · default: 1Speech speed multiplier (typical 0.5-2.0; provider-specific)
hd_qualityboolean · default: falseSwitch to a higher-fidelity model when the provider supports it
voice_settingsobjectFine-tuning. Supported keys (provider-dependent): stability (0-1), similarity_boost (0-1), speed, volume (0.1-10), pitch (-12 to 12), emotion.
clone_from_audiostringURL to an audio file. When set, the voice is cloned from this sample for this request before synthesis. Available on ElevenLabs and MiniMax.
clone_from_videostringURL to a video file. The audio track is extracted, then cloned.
keep_source_fileboolean · default: falseKeep your uploaded clone source after processing. By default it's deleted once the request completes.
auto_delete_after_hoursinteger · default: 24When keep_source_file is true, delete the source automatically after this many hours.
cinema8_envstringCinema8 environment hint (PROD or STAGING) — only relevant for Cinema8 media URLs.
data.urlstringURL to the generated .mp3 file — download or stream from here.
data.keystringStorage object key for the audio file.
data.folder_idstringIdentifier for the storage folder the audio was saved under.
data.providerstringThe provider that produced the audio.
data.voice_idstringThe voice ID that was actually used.
data.cloned_voice_idstringWhen you used clone_from_audio or clone_from_video, the cloned voice's ID. null otherwise.
data.duration_secondsnumberAudio duration in seconds — this is what you're billed on.
usage.unitsnumberSame as data.duration_seconds. cost.units mirrors this value.
Per-second billing on synthesized audio. Browsing voices is free.
| Service | Unit | Price |
|---|---|---|
| Text to Speech | second | 4 credits/second |
| List Voices | item | Free |
Billing is based on generated audio duration (usage.units = duration_seconds), not input text length.
A typical 1-minute narration costs approximately 60 credits.
Voice cloning via clone_from_audio/clone_from_video is included — no separate cloning charge.