# Scribe One — speech to text

> Scribe One is eroq's transcription model: automatic language detection, audio up to 8 MB per request, 5 credits flat. Built for voice messages and call modes in character products.

Scribe One turns audio into text for a flat 5 credits per request: voice messages, call-mode turns, short recordings up to 8 MB. Language is detected automatically. Paired with Voice One or Voice Turbo it closes the loop for spoken conversations with a character.

## At a glance

| | |
| --- | --- |
| Price | 5 credits / request |
| Input | Audio file up to 8 MB (≈ 8 minutes compressed) |
| Languages | Automatic detection |
| Endpoint | POST /v1/audio/transcriptions |
| Access | Every account |

## A call mode in three calls

Record the user, transcribe with Scribe One, answer with RP+ and speak the reply with Voice Turbo. Three requests, one key, one credit balance — the pattern behind voice modes in companion apps and interactive fiction.

## FAQ

### What audio formats does Scribe One accept?

Common compressed formats (mp3, m4a, ogg, wav) up to 8 MB per request, sent as multipart form data.

### Is it priced per minute?

No — a flat 5 credits per request, whatever the length under the 8 MB cap.

## Related

- https://eroq.ai/docs/transcriptions — API reference
- https://eroq.ai/models/eroq-voice-turbo — Voice Turbo
- https://eroq.ai/use-cases/ai-companion-apps — Companion apps

---

This page as HTML: https://eroq.ai/models/eroq-scribe-one · Studio: https://eroq.ai/studio · Docs: https://eroq.ai/docs · Machine index: https://eroq.ai/llms.txt