Awtomatikong salin · Orihinal sa Ingles

Voxtral TTS

Voxtral 4B TTS 2603 ay lokal na tumatakbo sa Apple Silicon sa pamamagitan ng Voxtral TTS module: 20 preseted voices, siyam na wika, at 24 kHz mono speech.

Utos

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

Nakatakdang mga tinig

`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`

Mga Hangganan at Pagmamay - ari

Ang mga preset voices lamang; reference-audio cloning ay hindi makuha. Ang audio output ay dumadaloy sa SDK, CLI, WebSocket at HTTP PCM. Isumite ang kumpletong input text up front. Ang mga model weights ay gumagamit ng CC BY-NC 4.0.

Ang bawat kahilingan ay pagmamay-ari ng cache ng wika, codec history at random seed.Ang Calleback delivery ay naghihintay sa mamimili. Ang default figure ay 3 frames (240 ms of audio), at ang limitasyon ay 500 frames (40 segundo).Ang isang diffing failure ay maaaring sumunod sa mas naunang audio cluster; ang normal na pagkumpleto lamang ay naglalabas ng isang pangwakas na marker.

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

MLX 4-bit · Maling Akala · MGA VoxtralTTE