MetaVoice-1B
The facts below are what the maker publishes about MetaVoice-1B, checked and dated. Our editors have not written about it yet.
The facts
Each one read from MetaVoice-1B's own pages, with the date it was checked.
What it does
- Access options
- The page lists local use, Hugging Face, and a Google Colab demo as ways to use the model.checked 12 Oct 2026
- API
- The inference server provides API definitions at its /docs route.checked 13 Oct 2026
- API access
- Yeschecked 20 Sep 2026
- Commercial use
- Yeschecked 20 Sep 2026
- Dataset caveat
- The fine-tuning workflow does not check for overlap between training and validation datasets, so the page advises keeping them disjoint.checked 12 Oct 2026
- Deployment
- The project provides a web UI and inference server that can be deployed on AWS, GCP, or Azure.checked 12 Oct 2026
- Expressive speech
- The project prioritizes emotional speech rhythm and tone in English.checked 12 Oct 2026
- Fine-tuning
- The project supports fine-tuning the first-stage language model and describes cross-lingual voice cloning with fine-tuning.checked 13 Oct 2026
- Fine-tuning data limit
- The fine-tuning instructions say the project does not check for overlap between training and validation datasets.checked 13 Oct 2026
- Hardware requirement
- Installation requires a GPU with at least 12 GB of VRAM, Python 3.10 to below 3.12, and pipx.checked 12 Oct 2026
- Hardware requirements
- Installation prerequisites include a GPU with at least 12 GB of VRAM and Python 3.10 or later but earlier than 3.12.checked 13 Oct 2026
- Integrations
- The README describes an optional Weights & Biases integration for logging fine-tuning runs.checked 13 Oct 2026
- License
- The model is released under the Apache 2.0 license, which the project describes as usable without restrictions.checked 13 Oct 2026
- Performance
- After compilation, the synthesise API runs faster than real time on Ampere, Ada-Lovelace, and Hopper GPUs, according to the project page.checked 12 Oct 2026
- Purpose
- MetaVoice-1B is a 1.2-billion-parameter text-to-speech model trained on 100,000 hours of speech.checked 13 Oct 2026
- Quantization limit
- Experimental int4 and int8 inference can run faster, but the project says this degrades audio quality.checked 13 Oct 2026
- Speech style
- The model is designed for emotional speech rhythm and tone in English.checked 13 Oct 2026
- Startup time
- The reference inference script takes 30 to 90 seconds to start, depending on hardware, because it compiles the model.checked 12 Oct 2026
- Support
- The repository directs users to its active GitHub issues for contributions and discussion.checked 13 Oct 2026
- Tracking integration
- The project offers optional Weights & Biases logging, enabled through configuration and installing the relevant dependencies.checked 12 Oct 2026
- Voice cloning
- It supports zero-shot cloning of American and British voices using 30 seconds of reference audio.checked 13 Oct 2026
- Ways to use
- The project documents a local reference implementation, a web UI, an inference server, Hugging Face usage, and a Google Colab demo.checked 13 Oct 2026
Money
- Free plan
- Yeschecked 20 Sep 2026
Where it runs
- Platforms
- web,api,self_hostedchecked 20 Sep 2026
Alternatives to MetaVoice-1B
- 1
iSpeech
Free planOpen sourceAndroidiOSWebAPI8.5Free plan - 2
Voicebox
Free planOpen sourceAndroidiOSWebAPI8.5From $12 a year - 3
Resemble AI
Free planAndroidiOSWebAPI8.3From $280 a month - 4
Speechify
Free planAndroidiOSWebAPI8.3From $10 a month - 5
Voice.ai
Free planAndroidiOSWebAPI8.3From $5 a month - 6
Fliki
Free planAndroidiOSWebAPI8.2From $28 a month