waven.ai
← Home

Credits and licenses

This service uses open-source text-to-speech, speech-to-text, speaker diarization, and language models. Below are the models available, their sources, and their license terms. If you use generated audio or transcripts commercially, check the license for the model that produced them — the model name is returned in every API response.

OmniVoice

Apache 2.0 permits commercial use with attribution. Include a notice that the audio was generated using OmniVoice when required by your use case.

Kokoro 82M

Apache 2.0 permits commercial use with attribution. Kokoro powers the fast narration tier; the OmniVoice tier handles voice-cloned output.

Parakeet-TDT 0.6B v2

English batch transcription. CC-BY-4.0 permits commercial use but requires attribution — credit NVIDIA when you publish transcripts produced by this model.

Voxtral-Mini-3B-2507

Multilingual transcription (streaming and batch). Apache 2.0 permits commercial use with attribution.

pyannote speaker-diarization-3.1

Speaker labels (diarization) on transcripts. MIT permits commercial use; cite the pyannote.audio technical reports as requested on the model card.

Qwen3-4B

Transcript cleanup and audio intelligence (summaries, sentiment, topics, entities) on transcription jobs. Apache 2.0 permits commercial use with attribution.

Additional models

When new models are added to the platform, their licenses and attribution requirements will be listed here. Check the model name shown in the API response to identify which license applies to your output, and consult the linked source for the current terms.

Terms of Service · Privacy Policy