Terjemahan mesin · Versi asli bahasa Inggris

Voxtral TTS

Voxtral 4B TTS 2603 berjalan lokal pada Apple Silicon melalui modul Voxtral TTS: 20 preset suara, sembilan bahasa, dan 24 KHz mono speech.

Baris perintah

speech speak "Hello, welcome to Voxtral." --engine voxtral --voxtral-voice casual_male --stream --voxtral-chunk-frames 3 -o hello.wav
speech speak --engine voxtral --list-speakers
speech speak "Bonjour." --engine voxtral --voxtral-voice fr_female --voxtral-max-frames 500 --voxtral-seed 0 -o bonjour.wav

Swift API

import VoxtralTTS

let model = try await VoxtralTTSModel.fromPretrained()
for try await chunk in model.generateStream(text: "Hello, welcome to Voxtral.", voice: "casual_male") {
    print(chunk.frameIndex, chunk.samples.count, chunk.isFinal)
}

Server API

curl http://localhost:8080/v1/audio/speech \
  -H 'Content-Type: application/json' \
  -d '{"model":"voxtral-tts","input":"Hello, welcome to Voxtral.","voice":"casual_male","response_format":"wav"}' \
  -o hello.wav

/v1/realtime: session.update → {"model":"voxtral-tts","voice":"casual_male"}; response.create → {"instructions":"Hello, welcome to Voxtral."}.

curl http://localhost:8080/v1/audio/speech \
  -H 'Content-Type: application/json' \
  -d '{"model":"voxtral-tts","input":"Hello, welcome to Voxtral.","voice":"casual_male","stream":true,"response_format":"pcm"}' \
  -o hello.pcm

Preset suara

`en`	`casual_male, casual_female, cheerful_female, neutral_male, neutral_female`
`fr`	`fr_male, fr_female`
`es`	`es_male, es_female`
`de`	`de_male, de_female`
`it`	`it_male, it_female`
`pt`	`pt_male, pt_female`
`nl`	`nl_male, nl_female`
`ar`	`ar_male`
`hi`	`hi_male, hi_female`

Batas dan kepemilikan

Hanya preset suara; reference-audio cloning tidak tersedia. Keluaran stream audio melalui SDK, CLI, WebSocket dan HTTP PCM. Pasokan teks masukan lengkap di depan. Model bobot menggunakan CC BY- NC 4.0.

Setiap permintaan memiliki cache bahasa, riwayat codec dan benih acak. Kiriman callback menunggu konsumen. Potongan baku adalah 3 frame (240 m audio), dan limit adalah 500 frame (40 detik). Kegagalan streaming dapat mengikuti potongan audio sebelumnya; hanya penyelesaian normal emits penanda akhir.

VoxtralTTSOptions(streamingChunkFrames: 3) · generateStreaming(onChunk:) · AudioChunk.isFinal.

MLX 4- bit · Mistral · VoxtralTS