Oriel — AI & technology worth noticing, every day.

BreezeBlue/Breeze-TTS-2

AI model · https://huggingface.co/BreezeBlue/Breeze-TTS-2

First seen by Oriel on .

AI summary

Breeze TTS 2 is an open-weight, bilingual text-to-speech model for real-time interaction, offering voice cloning, natural-language voice design and direction, and streaming inference.

  • The model supports English and Chinese speech, reference-based voice cloning, reference-free voice design, and voice direction for tone, emotion, pace, and delivery.
  • Inline text markers add vocal events, such as (laugh) in English and [笑] in Chinese.
  • On an NVIDIA H100, the warmed-up fast path reports under 40 ms time to first audio and a 0.32 real-time factor; eager inference uses about 7.7 GiB of GPU memory.
  • The streaming API serves single-concurrency requests and returns mono 24 kHz signed 16-bit little-endian PCM; the fast path requires more GPU memory and cold-start time.
  • Source code is Apache 2.0, but model weights and self-hosted outputs are limited to research and non-commercial use; commercial rights apply only to hosted outputs under an active paid subscription.

#text-to-speech #voice cloning #speech synthesis #gpu inference

Generated by gpt-5.6-sol. Check the original for details.

Metric trends

Last 30 days · Up to 120 recorded samples per source and metric, with reposts kept separate · First and latest retained

Likes

Total likes reported by the source.

  • Hugging Face Models: 909, latest observation ; +26 over observed period since

Downloads (30d)

Rolling 30-day downloads. A change in this value is not new downloads during the observed period.

  • Hugging Face Models: 21,483, latest observation ; +1,569 over observed period since

Aliases

SourceExternal IDURL
Hugging Face ModelsBreezeBlue/Breeze-TTS-2https://huggingface.co/BreezeBlue/Breeze-TTS-2 Read discussion