Skip to content
Audio

tiny-voca-ca

compact Catalan text-to-speech for the browser

2 min read

This compact 22 MB text-to-speech (TTS) engine generates Central Catalan speech directly in the browser. Its architecture splits into two ONNX files—one handling phonemes and acoustic features, the other generating the waveform. Built in Rust and compiled to WebAssembly, it keeps all user data completely on-device. It includes two professional voices from the FestCat dataset: female (ona) and male (pau).

Key features

  • Fully offline in the browser: runs serverless via WebAssembly and the tract engine.
  • Ultralightweight: weighs just 22 MB total, split into 7 MB and 15.5 MB files.
  • Two voices: supports ona and pau with 24 kHz mono audio output.
  • Fast inference: runs 6x faster than real time on a single thread on a Poco X6 smartphone.

What you can do

  • Web app integration: run local speech synthesis directly on web pages or inside a Web Worker.
  • Mobile web tools: power lightweight offline assistants in phone and tablet browsers.
  • Catalan voiceovers: generate speech on the fly without sending audio or text to remote servers.

Example

import { TinyVoca } from './tiny-voca.js';

const tts = await TinyVoca.load({ model: '[https://huggingface.co/ttzzs/tiny-voca-ca/resolve/main](https://huggingface.co/ttzzs/tiny-voca-ca/resolve/main)', language: 'ca' });
const audio = tts.speak('Bon dia, com va?', { voice: 'pau' });
Read moreDetails
Benchmarks: achieves a 14.8% WER on the test set (comparable to the original speaker recordings at 17.0%).
Limitations: struggles with acronyms, rare abbreviations, and question intonation; supports Central Catalan only.
License: CC BY-SA 4.0 (based on the FestCat dataset and the tiny-voca architecture).
Discuss Telegram 0

Will it run?

Working out the memory needs

The bot is still collecting this model's weight sizes; the estimate will appear here in a few minutes. Meanwhile you can rent a GPU with headroom.

Massed ComputeOpen Massed Compute

Run it on Massed Compute

The script downloads the model to the rented machine. How to run it is in the model card.

  1. 1. Rent a machine with a fitting GPU

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    # ttzzs/tiny-voca-ca
    # ===== Tokens: paste yours between the quotes =====
    export HF_TOKEN=""        # huggingface.co/settings/tokens, for gated models
    # ==========================================
    mkdir -p ~/models
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv tool install "huggingface_hub[cli]"
    hf download ttzzs/tiny-voca-ca --local-dir ~/models/tiny-voca-ca
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to tiny-voca-ca
  3. 3. Use it

    The model is in ~/models. How to run it is in the model card (link above).

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (2)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

Model family

New posts about versions of tiny-voca-ca (quants, finetunes, LoRAs) will show up under the bell.

No known relatives of this model yet. All derivatives on Hugging Face

About the model

ttzzs/tiny-voca-ca
Author
ttzzs
Category
Audio
Released
Posted

On NeyroLearn

Hugging Face

00

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments