Public API + MCP

Audio intelligence for your applications

Transcription, AI Notes and chat over any audio, accessible from your backend (REST API) or from Claude / ChatGPT / Cursor / Gemini (MCP server). Pay-per-credit, no commitment, data in Europe.

Create API key Connect via MCP

Real Spanish transcription

Transcription with AssemblyAI Universal-3.5 Pro, with diarization, native code-switching across languages and LATAM dialects. Quality superior to anglo-centric APIs.

Structured AI Notes

14 templates (minutes, sales, medical, legal, podcast…) with summaries, action items, decisions, mind map.

Chat over your audio

RAG with local BGE-M3 embeddings. Ask your recordings; answers cite timestamps with clickable seek.

REST API quickstart

4 steps to start integrating AudioMap from your backend.

These are the routes the server publishes right now, read from its own OpenAPI. They authenticate with the same key as MCP.

POST	/v1/notes	Import from a public link, or open a direct upload
GET	/v1/notes	List notes, newest first
GET	/v1/notes/{id}	Get one note with its full text, analyses and outline
POST	/v1/notes/{id}/finalize	Close a direct upload and start processing
GET	/v1/notes/{id}/transcript	Transcript with segments, speakers and word timings
GET	/v1/notes/{id}/transcript/translation	State and segments of the transcript translation
POST	/v1/notes/{id}/transcript/translation	Translate the transcript into a language
GET	/v1/notes/{id}/analysis	Read the generated analyses of a note
POST	/v1/notes/{id}/analysis	Generate or regenerate an analysis template
GET	/v1/notes/{id}/outline	Chapters of the note with timestamps
GET	/v1/notes/{id}/speakers	Speakers of the note with their names
POST	/v1/notes/{id}/speakers/suggest	Infer the speaker names from content and metadata
PATCH	/v1/notes/{id}/speakers/{label}	Name a speaker, or confirm its suggested name
GET	/v1/voice-profiles	Voice profiles of the user and the state of the biometric consent
POST	/v1/notes/{id}/ask	Ask a question about one note
POST	/v1/ask	Ask a question across every note you own
POST	/v1/notes/{noteId}/deliverables	Write a document from the note
GET	/v1/notes/{noteId}/deliverables	List the deliverables generated for a note

Read the API documentation →OpenAPI as JSON

Create an API key

Log into your account and generate a key at https://audiomap.ai/en/dashboard/api-keys

Upload audio for transcription

# Returns 202 with noteId, transcription is async
curl -X POST https://api.audiomap.ai/v1/notes \
  -H "Authorization: Bearer am_live_xxxxxxxxxxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "audioUrl": "https://example.com/meeting.mp3",
    "language": "es-ES"
  }'

# 202 Accepted
{
  "noteId": "01J...",
  "status": "processing"
}

Cost: 50 credits per minute of audio. If you exceed your monthly quota, returns 402.

Polling: get the result

curl https://api.audiomap.ai/v1/notes/01J... \
  -H "Authorization: Bearer am_live_xxxxxxxxxxxx"

# 200 OK
{
  "id": "01J...",
  "status": "ready",
  "transcript": { "fullText": "...", "languagesDetected": ["es"] },
  "aiNotes": [{ "summary": "...", "actionItems": [...] }],
  "outline": [{ "startMs": 0, "endMs": 60000, "headline": "..." }]
}

Webhooks (optional)

Configure outbound webhooks from the dashboard, under Integrations: AudioMap calls your endpoint when transcription completes, HMAC-signed and with retries. Available on the Power plan.

MCP quickstart

Connect AudioMap to Claude, ChatGPT, Cursor or Gemini in 30 seconds. Your AI client accesses your notes, transcripts and AI Notes directly.

Claude

Compatible

ChatGPT

Compatible

Cursor

Compatible

Gemini CLI

Compatible

OAuth 2.1 (recommended for IA clients)

Claude, ChatGPT, Cursor and VS Code discover OAuth automatically from the metadata endpoint and register themselves (dynamic client registration). Your users paste https://mcp.audiomap.ai, approve consent once and connect: no API key copy-paste.

# Protected resource (RFC 9728) and authorization server (RFC 8414)
curl https://mcp.audiomap.ai/.well-known/oauth-protected-resource
curl https://api.audiomap.ai/.well-known/oauth-authorization-server

Supported grant types: authorization_code + refresh_token · PKCE S256 · Access token TTL 1h · Refresh 30d · Hard expiry 90d.

Configure your MCP client

{
  "mcpServers": {
    "audiomap": {
      "transport": "streamable-http",
      "url": "https://mcp.audiomap.ai",
      "headers": {
        "Authorization": "Bearer am_live_xxxxxxxxxxxx"
      }
    }
  }
}

Paste this into your MCP client's config (Claude Desktop config, Cursor settings, ChatGPT Developer Mode…). Replace am_live_... with your API key.

Ask your AI

Once connected, your AI can use the tools without further config. Examples:

"Search my notes for what we said about last month's contract"
"Summarize the last client meeting and give me the action items"
"At what point in the May 5th audio did we discuss pricing?"

Available tools

All 15 tools are live in production. Write tools deduct credits according to your plan.

Tool	Description	Plan
search_notes	Search notes by text in title/description/transcript	Free
list_recent_notes	List the most recent notes	Free
get_note	Returns a complete note with AI Notes	Free
get_transcript_segments	Transcript segments with timestamps and speakers	Trimmed on the free plan
get_transcript_translation	Transcript translated into another language, per segment and cached.	Trimmed on the free plan
list_folders	List folders with note count	Free
list_highlights	Highlights from a note with timestamp	Free
get_outline	Auto-generated outline/chapters of a note	Free
list_markers	List your markers (points and segments) for a note	Free
get_speaker_timeline	Who spoke when in a note: turns per speaker and talk time	Free
import_from_url	Imports an audio or video from a link (YouTube, podcast RSS or Apple, direct URL) and transcribes it	Uses credits
wait_for_transcription	Poll until transcription completes	Free
generate_ai_notes	Generate AI Notes with a specific template (sales, legal, medical…)	Uses credits
generate_outline	Generate outline/chapters with timestamps	Uses credits
replace_in_transcript	Literal search & replace over the full transcript	Free
suggest_speaker_names	Suggests each speaker's name from what is said, the title and the channel	Free
rename_speaker	Names a speaker or confirms the suggested name	Free
list_voice_profiles	Voice profiles of the user and the state of the biometric consent	Free
founder_status	Your founder programme status and commitment	Free
ask_note	Ask a natural-language question about a note (RAG with citations)	Uses credits
extract_clip	Extract an audio clip between two timestamps	Uses credits

Public demo, no signup

Want to try AudioMap MCP before signing up? Connect the demo endpoint to your AI client and play with a sample dataset. 10 requests per 10 minutes per IP, read-only tools.

{
  "mcpServers": {
    "audiomap-demo": {
      "transport": "streamable-http",
      "url": "https://mcp.audiomap.ai/demo"
    }
  }
}

Demo dataset only, no access to your personal library. To access your real audio, sign up and create an API key.

Credits: how they work

The Pro and Power plans include monthly credits consumed when using the API or MCP. Read operations are free.

Credits per plan

Free (€0)200 / mo
Personal (€9)1,000 / mo
Pro (€19)5,000 / mo
Power (€39)25,000 / mo

Cost per operation

Transcribe 1 minute of audio50
Generate AI Notes100
Generate outline/chapters50
1 chat message10
Extract MP3 clip5
Search / list / read (always free)0

No overage. When the month's credits run out, calls return 402 until the next cycle or an upgrade. Unused credits don't roll over.

Start integrating in 5 minutes

Create your API key, copy the curl example or the MCP JSON, and start calling. The MCP server works on every plan, with each plan’s credits; the REST API requires the Pro plan or above.

Create API key

Need help? hola@audiomap.ai