For teams
Model access for the research groups and teams I work with, on a platform I operate.
Mainstream models, one API
Mainstream models are called through one OpenAI-compatible /v1; existing
SDKs and tools need only a new base URL and key.
Subscription allowance
Each member has their own key with a weekly allowance that renews every week. Usage is always visible on my.soundadam.com.
Team management
Members join by invitation link with one GitHub sign-in; granting and revoking take effect at once. Creating a key, setup notes and usage are self-service in the console.
Speech API
The same key and the same /v1 as the models:
| Model | Does | Endpoint |
|---|---|---|
qwen3-asr | Transcription, files up to 50 MB (about 26 minutes of 16 kHz mono wav) | /v1/audio/transcriptions |
qwen3-asr-aligned | Transcription with word timestamps | /v1/audio/transcriptions |
qwen3-tts | Speech synthesis in the model’s built-in voices, with an optional style prompt | /v1/audio/speech |
sensevoice-small | A small transcription model on CPU | /v1/audio/transcriptions |
kokoro | A small synthesis model on CPU | /v1/audio/speech |
Getting access
Write to hello@soundadam.com with your team’s size and what you need it for. I reply with an invitation link.