> ## Documentation Index
> Fetch the complete documentation index at: https://docs.ironlabs.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Voice input
> Speak your prompts. Foundry transcribes and sends.
Click the mic in the composer, speak, release. Foundry records, transcribes with Whisper, and drops the text into the input — you can edit before sending, or just hit Enter.
**Screenshot coming.** Mic button + recording state.
## When to use it
* Drafting long prompts where typing slows you down.
* Hands-free moments (cooking, driving, accessibility needs).
* Capturing a quick thought before it slips.
## How recording works
| Spec | Limit |
| ---------------------- | --------------------------------------------------------------------- |
| **Max recording time** | 5 minutes per take |
| **Max file size** | 5 MB |
| **Supported formats** | WebM, MP4, MPEG, WAV, OGG, M4A — Foundry picks based on your browser. |
| **Language** | 99+ languages auto-detected by Whisper. No language picker needed. |
If you go over the 5-minute or 5 MB limit, Foundry stops recording and uses what it captured up to that point.
## Tips
* **Punctuate verbally** — say "comma" or "new line" if you need them. Whisper handles natural pauses well, but explicit punctuation cleans up code-heavy prompts.
* **Edit before sending.** The transcript drops into the composer; tweak it, then Enter.
## Limits
* The first time you click the mic, your browser asks for microphone permission. If you deny it, the mic stays disabled — re-enable in your browser's site settings.
* Whisper is good at most accents and languages, but technical jargon and proper nouns sometimes need correction.
* No real-time transcription — Foundry transcribes when you release the mic.
## Related
Faster than the mouse.
Get pinged when long replies finish.