---
title: AI and speech tools
description: Configure optional AI providers for subtitle translation, formatting, file naming, and speech transcription.
---

mpvRx can connect to supported AI and speech providers for optional workflows. These features are not required for normal playback.

## AI providers

Supported integrations include OpenAI, Anthropic, Groq, OpenRouter, Together, and OpenCode Zen. Available models, pricing, retention, rate limits, and regional access are controlled by each provider.

Possible workflows include:

- Subtitle translation with a custom instruction
- Subtitle formatting cleanup
- Structured file-renaming suggestions
- Speech-to-subtitle generation through cloud or offline Whisper paths

## Configure safely

1. **Choose a provider**

    Review its privacy policy, pricing, and model support before creating credentials.

2. **Add credentials in mpvRx**

    Store the API key only in the provider setting. Do not embed it in a script, subtitle, filename, or issue report.

3. **Test with non-sensitive content**

    Start with a short file or subtitle and verify language, formatting, and cost behavior.

4. **Review every result**

    Generated text and filenames can be wrong. Keep the original material until the output is verified.

:::warning[Data leaves the device]
Cloud providers receive the text or audio needed for the selected operation. Offline Whisper avoids a cloud transcription request but still requires device resources and model support.
:::

## Troubleshooting

Provider errors commonly come from an invalid key, unavailable model, unsupported endpoint, quota, rate limiting, or a response format the selected workflow cannot parse. Record the provider and model, but redact keys and private content when asking for support.