# Arch Tools – Audio Transcription

> Arch Tools – Audio Transcription is a paid API for AI agents from archtools.dev, paid per call via x402, $0.03/call, status healthy (last checked 2026-09-15, last successful call 2026-09-04).

Transcribes an audio file from a public URL into text, with optional language hint and context prompt for improved accuracy

## Facts

- Endpoint: POST https://archtools.dev/v1/tools/transcribe-audio
- Price: $0.03/call
- Payment: x402
- Status: healthy
- Last checked: 2026-09-15
- Last successful call: 2026-09-04
- Success rate: 100% of calls made through Zero
- Activations on Zero: 21
- Tags: x402
- Canonical page: https://www.zero.xyz/c/arch-tools-audio-transcription-a5a512d9
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_xzSn43TZo2Ly48u12CbYS

Status and success rate cover calls made through Zero and Zero's own probes. Third-party monitors may report differently.

## How to call it through Zero

Zero handles the 402 payment challenge and records the run. With the Zero CLI installed (`npm i -g @zeroxyz/cli`):

```sh
zero fetch --capability arch-tools-audio-transcription-a5a512d9 -d '<json body>'
```

Example prompt: Can you transcribe the audio at https://example.com/meeting-recording.mp3 for me? It's an English interview, and you can use the hint 'quarterly earnings discussion' for better accuracy.

## When to prefer this

Choose this endpoint when you have a public URL pointing to an audio file (MP3, WAV, M4A, or OGG) and need a text transcript quickly, especially within an automated agent workflow that already uses x402 micropayment infrastructure. It is particularly useful when you can supply a context hint to improve domain-specific accuracy. Prefer this over generic STT APIs if you are already operating in the Arch Tools ecosystem or need pay-per-call pricing without subscription commitments.

## Known failure modes

- Audio URL is not publicly accessible or returns a non-200 response — transcription fails
- Unsupported audio format (formats other than MP3, WAV, M4A, OGG) — request rejected
- Audio file too large or too long — may time out or return an error
- Incorrect language code provided — may reduce accuracy or fail
- Payment not completed via x402/USDC on Base — request rejected with 402
- Audio contains no intelligible speech — returns empty or minimal transcript

## How this service works

Transcribe audio to text in 100+ languages via Whisper. Send a file URL, get an accurate transcript back in one call. Pay per call with USDC (x402) or credits - archtools.dev

## Output

A JSON object containing the transcribed text extracted from the audio file at the provided URL. The transcript reflects the spoken content, potentially improved by any language code or context prompt supplied in the request.

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "properties": {
  "prompt": {
   "type": "string",
   "description": "Context hint for better accuracy"
  },
  "language": {
   "type": "string",
   "description": "ISO-639-1 language code (auto-detected if omitted)"
  },
  "audio_url": {
   "type": "string",
   "description": "Public URL of the audio file (MP3, WAV, M4A, OGG)"
  }
 }
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/arch-tools-audio-transcription-a5a512d9/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from archtools.dev](https://www.zero.xyz/host/archtools.dev/llms.txt)