- Generate high-quality audio from text, locally and in multiple languages.
- Produce downloadable MP3 files.
- Choose the right voice and language.
Key concepts
This is pure TTS (text-to-speech) — no microphone, no language model: just text in and audio out.
Which tools to use (2026)
- Kokoro — lightweight, fast on CPU, good quality, multilingual. A great starting point.
- MagpieTTS — production-quality voices in 9 languages (including Spanish); works best with a GPU.
- F5-TTS — if you want to clone a specific voice from a sample.
Step by step
cd ~/proyectos-ia mkdir texto-a-audio cd texto-a-audio claude
Run it on your computer
npm install npm run dev
Paste a paragraph, choose language and voice, and generate the audio.
If something goes wrong
- Wrong-language voice — select the correct voice for that language; not every voice speaks every language.
- Cuts off on long texts — confirm the app splits and stitches chunks; if not, ask Claude Code to add that.
- Sounds metallic — try MagpieTTS for higher quality (works best with a GPU).
Practice challenge
Add a "podcast mode": have two different voices read a dialogue, alternating lines. Ideal for more engaging educational material.