How to use the Avalon ASR model

Avalon is Aqua’s speech recognition model built for real work: coding terms, AI jargon, company/product names, and even noisy environments. It’s more accurate than our default model and faster, so you make fewer corrections and ship work sooner.

Enable Avalon in Aqua Voice

Open Aqua Voice → Settings → Model
Select Avalon 1.5
You’re done. Avalon now powers dictation everywhere.

Use it anywhere you type

Put your cursor in any text field (email, docs, code editor, terminal, Slack, etc.).
Press your push-to-talk key (or click the mic), speak naturally, release to commit.

Where you’ll notice the biggest gains

Code & CLI: function/class names, file paths, flags, shell commands
AI & research: model names (e.g., “Llama 3.1,” “o3”), framework jargon, acronyms
Product/brand names: company and tool names that general ASR often mangles
Meetings & busy spaces: better resilience in noisy or overlapping speech

Troubleshooting

I can’t select Avalon in the model list.

Update Aqua to the latest version, then check the models available to your account in Settings → Model. Avalon 1.5 supports multiple languages, so you do not need to switch to English. See the language settings guide to choose your dictation language.

FAQ

Is Avalon faster or slower than the default model?

Avalon is faster. You’ll feel it in everyday use (fewer delays, fewer edits).

Is there an API?

Yes. See Avalon API (OpenAI-compatible): https://aquavoice.com/avalon-api

Is Avalon English-only?

No. Avalon 1.5 supports multiple languages. You can choose your dictation language in Settings → Language. See the language settings guide for details.

Introducing Avalon

Avalon on the OpenASR Leaderboard