CMF speech models
This Space separates text-to-speech (TTS) from speech-to-text (ASR). Only controls marked available execute today; planned ASR models are visible for transparent integration tracking, not as a simulated demo.
Choose a reviewed CMF TTS contract. The available Phonon model extracts assets locally and uses ptts; pending models stay visible for transparent release tracking but their synthesis button is disabled until an immutable CMF artifact is verified.
CMF artifact: infosave/phonon-40m-multilingual-cmf @ fcbc781f9108da6ea6c3a90b00e5c25875faac52 \nUpstream: gradium/phonon-40m-multilingual @ 1537ce46c64f42903e0d608b132ccb073dac46d2 \nThis is a CC BY-NC 4.0 derivative for non-commercial evaluation. Its declared TTS languages are English, Français, Deutsch, Español, Português.
Only listed model contracts can be selected; pending releases cannot run.
Selected model: Phonon 40M Multilingual (CMF)
Status: Available — verified CMF artifact + ptts runtime
Runnable languages: English (en), Français (fr), Deutsch (de), Español (es), Português (pt)
Choose the TTS model first; its selection narrows languages. Russian and other extended-language examples require Qwen3-TTS.
Selected language: English (en). The sample text above is ready to synthesize.
Temperature, seed, and CFG coefficient work with the default ptts runtime. Chunk limit is available only with an installed native CMF runtime. They are disabled for Qwen3-TTS CustomVoice because its reviewed UI exposes only text, language, and the fixed speaker catalog.
Available Phonon examples
| Text to synthesize | TTS language | Speech rate adjustment |
|---|
Qwen3-TTS CustomVoice examples
The following text is ready in the language selector when Qwen3-TTS is selected. Russian is intentionally available here through Qwen3-TTS, not through Phonon.
| Language | Code | Example text |
|---|---|---|
| 中文 | zh |
你好！这是 Qwen3-TTS 多语言语音合成示例。 |
| English | en |
Hello! This is a Qwen3-TTS CustomVoice synthesis example. |
| 日本語 | ja |
こんにちは。これは Qwen3-TTS の音声合成例です。 |
| 한국어 | ko |
안녕하세요. 이것은 Qwen3-TTS 음성 합성 예시입니다. |
| Deutsch | de |
Guten Tag! Dies ist ein Qwen3-TTS-Sprachsynthese-Beispiel. |
| Français | fr |
Bonjour ! Voici un exemple de synthèse vocale avec Qwen3-TTS. |
| Русский | ru |
Привет! Это пример синтеза речи с помощью Qwen3-TTS. |
| Português | pt |
Olá! Este é um exemplo de síntese de voz com Qwen3-TTS. |
| Español | es |
¡Hola! Este es un ejemplo de síntesis de voz con Qwen3-TTS. |
| Italiano | it |
Ciao! Questo è un esempio di sintesi vocale con Qwen3-TTS. |
| Speaker | Description | Native language |
|---|---|---|
Serena (serena) |
Warm, gentle young female voice | Chinese |
Vivian (vivian) |
Bright young female voice | Chinese |
Uncle Fu (uncle_fu) |
Seasoned male voice, mellow timbre | Chinese |
Ryan (ryan) |
Dynamic male voice with rhythm | English |
Aiden (aiden) |
Sunny American male voice | English |
Ono Anna (ono_anna) |
Playful Japanese female voice | Japanese |
Sohee (sohee) |
Warm Korean female voice | Korean |
Eric (eric) |
Lively Chengdu male voice | Chinese (Sichuan) |
Dylan (dylan) |
Youthful Beijing male voice | Chinese (Beijing) |
Do not submit confidential text. Outputs and optional uploads are temporary. Phonon supports only English, French, German, Spanish, and Portuguese; it accepts a precomputed voice latent when you have permission to use it. Qwen3-TTS CustomVoice adds Chinese, Japanese, Korean, Italian, and Russian through nine built-in speakers, using its separately published, verified CMF artifact.