Breeze TTS 2

audio_tokenizer/LICENSE · Apache-2.0 · Qwen/Qwen3-TTS-Tokenizer-12Hz

Apple M5 Pro · 48 GiB · TTFA / RTF / RSSS

2026-10-30 · en · 8bit=3.20 s · bf16=3.20 s · seed=42 · CFG=4 · firstChunkFrames=3 · chunkFrames=25 · MLX cache=2 GiB · warm=3

MLX	TTFA p50	RTF	RSS (GiB)	Pinakamataas na pisikal na bakas ng paa (GiB)
8bit	178 ms	0.535	3.92	7.23
bf16	272 ms	0.868	6.06	9.22

WAV=45 · distinct=11 · 8bit PCM SHA256=baseline · unit=1933 · skipped=42 · E2E=15 · Qwen3-ASR-0.6B-MLX-4bit

BF16 mga hakbang ngayon 0.868 RTF sa maikling pangungusap (baseline: 1.258) at 0.853 tungkol sa 9.20 parapo (baseline: 1.280). Ang isang paunang balangkas ay nagbibigay ng unang PCM sa unang PC 138 Mga m. 11 Ang iba't ibang pangwakas na mga waveform ay gumawa ng sero na normal na pagbabago WER/CER sa ganang sarili ASR; 8-bit default-chunk output ay byte-identical sa baseline. BF16 Ang pag-ikot ay maaaring magbago ng audio na may seed. set SPEECH_SWIFT_BREEZE_GEMV=0 bago magkarga upang ihambing ang orihinal na daang usli.

Para sa mas mababang playback startup, gumamit ng isang paunang frame, dalawang advide-up frames, at isang 1 GiB MLX cache budget. Ang 9.04 s 8-bit na parapo ay umaabot sa tinatayang CLI prebuffer sa 190 ms, na walang ekstrang pag-erecording na kailangan sa pamamagitan ng mga naitalang crewnction comeback. Ito ay nag-referce ng callback at output-vice late.

2026-10-30 · 8bit · 9.04 s · firstChunkFrames=1 · chunkFrames=2 · MLX cache=1 GiB · PCM=103 ms · buffer=190 ms · RTF=0.761 · peak=6.10 GiB

speech speak "The quick brown fox jumps over the lazy dog." --engine breeze \
  --instruct "A clear, calm English narrator." --first-chunk-frames 1 \
  --chunk-frames 2 --breeze-cache-limit-mib 1024 --stream -o fast-start.wav

Ang TTFA ay nag-iindorso ng modelong loading at speaker playback. Results depende sa hardware, teksto, at setting. Ang kabahaging model policy caps reusable MLX cache sa mas maliit ng 2 GiB at isang-kapat ng pisikal na memorya, na nagpaparangal sa mas mahigpit na mga badyet sa aplikasyon. --breeze-cache-limit-mib ay maaaring magpababa sa prosesong ito-wide na kisame; 0 influes reusable caching. Live model tensors ay hindi limitado.

English/Chinese TTS na may disenyo ng boses, reference cloning at direksiyon, at streamed audio (MLX BF16/8-bit, 24 kHz; research/non-commercial lisensiya).

BreezeTTS · 24,000 Hz · 16 × 2,048 · 80 ms · MLX BF16 / 8-bit

Paggamit

speech speak "The quick brown fox jumps over the lazy dog." --engine breeze \
  --instruct "A clear, calm English narrator." --stream -o design.wav

speech speak "The weather is beautiful today." --engine breeze \
  --voice-sample reference.wav --breeze-reference-text "Exact transcript." \
  --breeze-seed 42 -o clone.wav

speech speak "We need to discuss what happened." --engine breeze \
  --voice-sample reference.wav --breeze-reference-text "Exact transcript." \
  --instruct "Speak slowly, with a serious tone." --stream -o directed.wav

Swift API

import BreezeTTS

let model = try await BreezeTTSModel.fromPretrained()
let samples = try await model.generate(
    text: "The quick brown fox jumps over the lazy dog.",
    instruction: "A clear, calm English narrator.")

try await model.generateStreaming(text: "Hello, world!", instruction: "A warm voice.") { chunk in
    print(chunk.sampleRate, chunk.samples.count, chunk.isFinal)
}

Mga Mapagpipilian

--breeze-model-id: aufklarer/Breeze-TTS-2-MLX-8bit
--breeze-bundle-dir
--breeze-cache-limit-mib: Optional process-wide reusable MLX cache budget sa MiB; 0 ventruits caching.
--breeze-reference-text
--breeze-temperature: 0.9
--breeze-top-k: 50
--breeze-top-p: 1
--breeze-cfg-scale: 4
--breeze-max-frames: 750
--breeze-seed: 42
--first-chunk-frames: 3
--chunk-frames: 25

Lisensiya

Hinango mula sa Breeze TTS 2 ni Breeze Blue at lisensiyado para sa pananaliksik at hindi-komersyal na paggamit lamang.

Pananaliksik sa Breeze Blue at Di-Commercial License · Breeze TTS 2 MLX 8-bit · Breeze TTS 2 MLX BF16

Pinagmulan at Pagkakumberte

speech-swift / BreezeTTTS · Transportasyon_breeze_mlx.py · WER / CER / RSS / TTFA / RTF