CoolFace
Modelpublic

drbaph/Breeze-TTS-2-comfyui

sourceHugging Faceotherupdated 27d agoView on Hugging Face
14likes1.5kdownloads
Model Card

Breeze TTS 2 — ComfyUI derivatives

Converted from BreezeBlue/Breeze-TTS-2 for the ComfyUI-Breeze-TTS-2 node pack.

Screenshot 2026-08-26 160830

Contents

FileDescription
Breeze-TTS-2-bf16.safetensorsThe official checkpoint merged from 2 shards into a single file (bit-exact, bf16, 6.49 GiB).
Breeze-TTS-2-int8-convrot.safetensorsAll 462 transformer projection linears INT8 ConvRot (backbone + text encoder + depth decoder), comfy-kitchen per-channel per-row scales. Smallest build.
Breeze-TTS-2-int8-hybrid.safetensorsBackbone + text encoder INT8 ConvRot; the depth decoder (the per-frame autoregressive hot loop) stays bf16. Recommended INT8 build: bf16-level speed with −27 % VRAM.
Breeze-TTS-2-int8-text-encoder.safetensorsText encoder only INT8; everything decode-time stays bf16.
audio_tokenizer/Unmodified Qwen3-TTS 12 Hz audio tokenizer (codec) used for reference-audio encoding and speech decoding.
quantization_report_*.jsonPer-tensor dequantization error statistics for each INT8 build.
benchmark_report.jsonRTF / VRAM / log-mel / Whisper-WER comparison across all four builds.

Benchmarks (RTX 5090, ComfyUI loader, sdpa, 24 kHz output)

Buildavg RTF (lower = faster)Peak VRAMWhisper WER vs prompt*
bf165.687.51 GiBbaseline
int8-convrot9.08 (+60 %)5.21 GiB (−31 %)matches bf16
int8-hybrid5.64 (parity)5.53 GiB (−27 %)matches bf16
int8-text-encoder5.346.82 GiB (−9 %)matches bf16

*All builds including bf16 share the same few ASR-level blurs on the clone prompt; design/direction prompts transcribe identically to bf16 on the hybrid builds.

The depth decoder runs 15 skinny single-token GEMMs per 12.5 Hz frame (≈1260 linear calls/frame), where int8 per-call overhead (quantize + dispatch, ~0.1 ms/call on Windows) outweighs the GEMM savings — that is why full INT8 regresses and the hybrid does not.

License

  • —Weights: the Breeze TTS 2 model weights, these derivative files, and self-hosted outputs are governed by the BreezeBlue Research and Non-Commercial License (see license_link). Commercial use requires written authorization from RESONIA, INC.
  • —Node-pack code: ComfyUI-Breeze-TTS-2 is Apache-2.0. The Breeze TTS source code is also Apache-2.0.