Request from Jatin. Source post: https://x.com/BreezeBlueX/status/2093370832454287496
Breeze TTS 2 is text-to-speech (open weights on Hugging Face as BreezeBlue/Breeze-TTS-2, inference at breezeblue-ai/breeze-tts). The gateway today is speech-to-text. This is a new surface, not another catalog ASR row.
Handy Breeze ASR (whisper.cpp Medium / Large v3 already in the catalog) is a different model family. Do not treat this request as "add another Breeze STT pin."
Known constraints if anyone picks it up:
- Weights, derivatives, and self-hosted outputs are under the BreezeBlue Research and Non-Commercial License (Apache 2.0 is only the inference code).
- Official run path wants a CUDA NVIDIA GPU (about 8 to 14 GB depending on eager vs fast).
- Product site is https://breezeblue.ai
Do not ship this as on-device gateway mode. Audio for TTS would still leave the client for a host we run.
This issue is the request only. No implementation attached.
Request from Jatin. Source post: https://x.com/BreezeBlueX/status/2093370832454287496
Breeze TTS 2 is text-to-speech (open weights on Hugging Face as BreezeBlue/Breeze-TTS-2, inference at breezeblue-ai/breeze-tts). The gateway today is speech-to-text. This is a new surface, not another catalog ASR row.
Handy Breeze ASR (whisper.cpp Medium / Large v3 already in the catalog) is a different model family. Do not treat this request as "add another Breeze STT pin."
Known constraints if anyone picks it up:
Do not ship this as on-device gateway mode. Audio for TTS would still leave the client for a host we run.
This issue is the request only. No implementation attached.