> For the complete documentation index, see [llms.txt](https://langtech-bsc.gitbook.io/applied-intelligence/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://langtech-bsc.gitbook.io/applied-intelligence/models/speech-models.md).

# Speech

Speech recognition (ASR) and text-to-speech (TTS) models for Catalan.

Our speech models cover both directions: **speech recognition** (audio → text) and **speech synthesis** (text → audio). They are trained on thousands of hours of Catalan speech, across all of its main accents.

## Speech recognition (ASR)

We recommend two ASR models because they solve different problems:

| Model                                                                                                                                                   | Output                       | Languages         | Size | Use it when…                                                                                                                            |
| ------------------------------------------------------------------------------------------------------------------------------------------------------- | ---------------------------- | ----------------- | ---- | --------------------------------------------------------------------------------------------------------------------------------------- |
| [**whisper-large-v3-ca-punctuated-3370h**](https://langtech-bsc.gitbook.io/applied-intelligence/models/speech-models/whisper-large-v3-ca-punctuated.md) | Punctuated, capitalized text | Catalan           | 1.5B | People will **read** the transcript: subtitles, minutes, documents                                                                      |
| [**stt\_ca-es\_conformer\_transducer\_large**](https://langtech-bsc.gitbook.io/applied-intelligence/models/speech-models/conformer-transducer-ca-es.md) | Lowercase, unpunctuated text | Catalan + Spanish | 120M | You need the **lowest WER**, for benchmarks and evaluations or for a downstream NLP pipeline. Also when audio mixes Catalan and Spanish |

{% hint style="info" %}
**Why two models?** Word error rate (WER) is computed on normalized text, so a model trained to output plain lowercase text scores better on it. Whisper's punctuation and casing make transcripts much more readable but don't count in WER. Choose by what you need: readable transcripts or the best evaluation score.
{% endhint %}

## Text-to-speech (TTS)

Both TTS models are **Matxa-TTS v2**, a fast, non-autoregressive model that reads raw text (graphemes). No phonemizer is needed.

| Model                                                                                                                                           | Voices                 | Accents                                               | License        | Use it when…                                                                    |
| ----------------------------------------------------------------------------------------------------------------------------------------------- | ---------------------- | ----------------------------------------------------- | -------------- | ------------------------------------------------------------------------------- |
| [**matxa-tts-v2-ca-central-graphemes**](https://langtech-bsc.gitbook.io/applied-intelligence/models/speech-models/matxa-tts-central.md)         | Central Catalan voices | Central                                               | GPL-3.0        | You need a standard Central Catalan voice, including for commercial use         |
| [**matxa-tts-v2-ca-multiaccent-graphemes**](https://langtech-bsc.gitbook.io/applied-intelligence/models/speech-models/matxa-tts-multiaccent.md) | 16 voices              | Balearic, Central, North-Western, Valencian, Northern | Non-commercial | You need regional accents or a choice of voices, for research or non-profit use |


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://langtech-bsc.gitbook.io/applied-intelligence/models/speech-models.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.
