audio-transcriber
Summary: Transcribe audio and video to text with Whisper — from files, microphone recordings, or base64 payloads — then export, ingest into a knowledge graph, or stream the results.
Transcribe files or microphone audio (
transcribe_audio) choosing model, backend (faster-whisper/openai-whisper), language, temperature, FP16, word timestamps, and export formats such as txt/srt.Translate speech to English by setting the task to
translateon the transcription tools.Transcribe verified base64 payloads (
transcribe_speech) using a sha256 digest, with language detection, segment/word timestamps, optional diarisation reporting, and typed refusal codes (digest mismatch, unsupported media, too large/long, transcription failure).Produce AudioSegment-style segments (
transcribe_media_transcribe_segments) that never raise, returning{"available": false, "error": ...}on failure.Ingest transcriptions into the knowledge graph (
audio_ingest_transcription), storing audio as an asset, text as a document, and Whisper segments as nodes.Export transcriptions to specified formats (
audio_transcriber_export).Record and manage audio streams: initiate (
audio_transcriber_initiate_stream), record for a duration (audio_transcriber_record), save frames to WAV (audio_transcriber_save_stream), and stop/close (audio_transcriber_stop_stream).Interact with a PersonaPlex server over WebSocket (
audio_transcriber_interact).Check engine health (
health_check) — readable Whisper backends, ffmpeg availability, default model and directory — without loading a model.
Provides native OpenTelemetry telemetry and tracing exports, enabling observability of the MCP server's operations.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@audio-transcriber transcribe the file meeting.mp3"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Audio Transcriber
CLI or API | MCP
Documentation — Installation, deployment, and usage across the CLI, Python API, official documentation.
Related MCP server: audio-transcription-mcp
Overview
Audio Transcriber is a production-grade Agent and Model Context Protocol (MCP) server designed to interface directly with Transcribe the operator's .wav .mp4 .mp3 .flac files to text or record the operator's own audio!.
Key Features
Consolidated Action-Routed MCP Tools: Minimizes token overhead and eliminates tool bloat in LLM contexts by grouping methods into optimized, togglable tool modules.
Enterprise-Grade Security: Complete support for Eunomia policies, OIDC token delegation, and granular execution context tracking.
Native Telemetry & Tracing: Out-of-the-box OpenTelemetry exports and native Langfuse tracing.
CLI or API
This agent wraps the Transcribe the operator's .wav .mp4 .mp3 .flac files to text or record the operator's own audio! API. The operator can interact with it programmatically or via its integrated execution entrypoints.
Detailed instructions on how to use the underlying API wrappers, extended schema bindings, and developer SDK references are maintained in docs/index.md.
MCP
This server use dynamic Action-Routed tools to optimize token overhead and maximize IDE compatibility.
Available MCP Tools
The table below is auto-generated from the live server — do not edit by hand.
Condensed action-routed tools (MCP_TOOL_MODE=condensed)
MCP Tool | Toggle Env Var | Description |
|
| Transcribe an audio file and natively ingest it into the knowledge graph. |
|
| |
|
| Transcribes audio from a provided file or by recording from the microphone. |
|
| Verify the digest, transcribe via Whisper, and return AudioSegment-shaped |
Verbose 1:1 API-mapped tools (MCP_TOOL_MODE=verbose or both)
MCP Tool | Toggle Env Var | Description |
|
| Export transcription to specified formats. |
|
| Initiate the audio input stream. |
|
| Interact with PersonaPlex server via WebSocket. |
|
| Record audio for a specified duration or until stopped. |
|
| Save the recorded frames to a WAV file. |
|
| Stop and close the audio stream. |
|
| Transcribe the audio file using the initialized backend. |
4 action-routed tool(s) · 7 verbose 1:1 tool(s). Each is enabled unless its <DOMAIN>TOOL toggle is set false; MCP_TOOL_MODE selects the surface (intent default — the six verb-tools, granular set loaded on demand · condensed action-routed · verbose 1:1 · both). Auto-generated — do not edit.
Detailed tool schemas, parameter shapes, and validation constraints are preserved in docs/usage.md.
Dynamic Tool Selection & Visibility
This MCP server supports dynamic toolset selection and visibility filtering at runtime. This allows the operator to restrict the set of exposed tools to prevent blowing up the LLM's context window.
The operator can configure tool filtering via multiple input channels:
CLI Arguments: Pass
--toolsor--toolsets(or their disabled counterparts--disabled-toolsand--disabled-toolsets) during startup.Environment Variables: Define standard environment variables:
MCP_ENABLED_TOOLS/MCP_DISABLED_TOOLSMCP_ENABLED_TAGS/MCP_DISABLED_TAGS
HTTP SSE Request Headers: Pass custom headers during transport initialization:
x-mcp-enabled-tools/x-mcp-disabled-toolsx-mcp-enabled-tags/x-mcp-disabled-tags
HTTP SSE Request Query Parameters: Append query parameters directly to the operator's transport connection URL:
?tools=tool1,tool2?tags=tag1
When query strings or parameters are provided, an LLM-free Knowledge Graph resolution layer (using DynamicToolOrchestrator) matches query intents against known tool tags, names, or descriptions, with safe fallback and automated 24-hour background cache refreshing.
MCP Configuration Examples
Install the connector-focused
[mcp]extra. Examples useaudio-transcriber[mcp]to add FastMCP / FastAPI throughscion-sdk; the required Scion SDK core still carriesmedulla[full]. The[agent-runtime]extra additionally enables model orchestration.
stdio Transport (local IDEs — Cursor, Claude Desktop, VS Code)
{
"mcpServers": {
"audio-transcriber-mcp": {
"command": "uvx",
"args": [
"--from",
"audio-transcriber[mcp]",
"audio-transcriber-mcp"
],
"env": {
"MCP_TOOL_MODE": "intent",
"AUDIO_PROCESSINGTOOL": "True",
"MEDIA_SIDECARTOOL": "True",
"MISCTOOL": "True",
"TRANSCRIBE_DIRECTORY": "/path/to/transcribe_directory",
"WHISPER_MODEL": "base"
}
}
}
}Runtime references require an alias-aware launcher such as Arbora. Other launchers must omit those entries and inject the resolved values through their own runtime secret boundary.
Streamable-HTTP Transport (networked / production)
{
"mcpServers": {
"audio-transcriber-mcp": {
"command": "uvx",
"args": [
"--from",
"audio-transcriber[mcp]",
"audio-transcriber-mcp",
"--transport",
"streamable-http",
"--port",
"8000"
],
"env": {
"TRANSPORT": "streamable-http",
"HOST": "127.0.0.1",
"PORT": "8000",
"MCP_TOOL_MODE": "intent",
"AUDIO_PROCESSINGTOOL": "True",
"MEDIA_SIDECARTOOL": "True",
"MISCTOOL": "True",
"TRANSCRIBE_DIRECTORY": "/path/to/transcribe_directory",
"WHISPER_MODEL": "base"
}
}
}
}Alternatively, connect to a pre-deployed Streamable-HTTP instance by url:
{
"mcpServers": {
"audio-transcriber-mcp": {
"url": "http://localhost:8000/audio-transcriber-mcp/mcp"
}
}
}Run a reviewed container image as a least-privilege stdio child (no listener or published port):
docker run -i --rm \
--read-only \
--cap-drop=ALL \
--security-opt=no-new-privileges \
--pids-limit=256 \
--tmpfs /tmp:rw,noexec,nosuid,nodev,size=64m \
-e TRANSPORT=stdio \
-e MCP_TOOL_MODE=intent \
-e AUDIO_PROCESSINGTOOL=True \
-e MEDIA_SIDECARTOOL=True \
-e MISCTOOL=True \
-e TRANSCRIBE_DIRECTORY=/path/to/transcribe_directory \
-e WHISPER_MODEL=base \
registry.example.invalid/audio-transcriber@sha256:<digest> audio-transcriber-mcpFor containerized network HTTP, supply an authenticated TLS ingress (or
direct server TLS), exact MCP_ALLOWED_HOSTS, and an exact trusted-proxy
CIDR policy through the operator-owned deployment profile. The generator
does not emit an unauthenticated non-loopback listener.
Auto-generated from the code-read env surface (MCP_TOOL_MODE + package vars) — do not edit.
Additional Deployment Options
audio-transcriber can run as a local stdio process or container, or behind a remote
network boundary. The
Deployment guide carries
the detailed transport contract.
Local container — launch a reviewed immutable image as a least-privilege stdio child with no listener or published port.
Remote URL — connect through an operator-supplied authenticated HTTPS ingress. Keep its URL, outbound identity references, trust profile, and exact
MCP_ALLOWED_HOSTSinAgentConfig.
Agent
Configure transcription (optional)
export WHISPER_MODEL="base" export TRANSCRIBE_DIRECTORY="/path/to/transcribe_directory"
Connector-focused MCP server (includes the shared graph engine)
uv pip install "audio-transcriber[mcp]"
Everything (development)
uv pip install "audio-transcriber[all]" # or: python -m pip install "audio-transcriber[all]"
### Container images (`:mcp`)
One multi-stage `docker/Dockerfile` builds the MCP-only image (`--target mcp`):
| Image tag | Build target | Contents | Entrypoint |
|-----------|--------------|----------|------------|
| `example/audio-transcriber:mcp` | `--target mcp` | `audio-transcriber[mcp]` — **connector-focused**, includes `medulla[full]`; no model-orchestration stack | `audio-transcriber-mcp` |
```bash
docker build --target mcp -t example/audio-transcriber:mcp docker/ # connector-focused MCP serveragent (immutable agent digest) with a co-located :mcp sidecar.
Knowledge-graph database (medulla)
Scion SDK core dependency (medulla[full]). The [mcp] extra keeps
deployments can use the bundled engine. For production or shared state, run
medulla as a dedicated database service and configure the runtime to use it.
Deployment recipes (single-node + Raft HA), connection configuration, and architecture
diagrams are documented in the
medulla deployment guide.
Repository Owners
Documentation
The complete documentation is published as the official documentation site and is the recommended reference for installation, deployment, and day-to-day operation.
Page | Contents |
pip, source, extras, prebuilt Docker image | |
run the MCP server and agent, Compose, Caddy + Technitium, env config | |
the MCP tool, the | |
capability summary and ecosystem role | |
concept registry ( |
Contribute
Contributions are welcome! Please ensure code quality by running local checks before submitting pull requests:
Format code using
ruff format .Lint code using
ruff check .Validate type-safety with
mypy .Ran test suites using
pytest
Deploy with scion-sdk-deployment
Provision this package with the consolidated scion-sdk-deployment
workflow. It selects an installed-package, editable-source, or immutable-container
path; records only runtime secret and TLS-profile references in AgentConfig; and
runs doctor, registration, policy, observability, and rollback gates. Ask your agent
to "deploy audio-transcriber with scion-sdk-deployment".
Install mode | Command |
Installed package |
|
Immutable container | deploy |
The repository embeds no deployment profile, credential value, certificate path, or
environment-specific endpoint. Supply those at runtime through AgentConfig and the
configured secret provider.
Governed capability contract
This package ships a compact canonical skill surface with specialist procedures
kept as referenced workflows. The current MCP tools, skill metadata,
connector_manifest.yml, ontology, mappings, shapes, fixtures, migrations,
tool-schema fingerprints, and certification metadata form one versioned
capability contract. Validate them together; do not rely on stale tool names or
historical per-task skill wrappers.
Runtime endpoints, credentials, certificate trust, tenant identity, retention, and observability policy are deployment inputs and are never packaged values. See Configuration, trust, and privacy before enabling a network transport, connector ingestion, Arbora delegation, or trace export.
Available Tools
12 toolsaudio_ingest_transcriptionAudio Ingest TranscriptionA
Transcribe an audio file and natively ingest it into the knowledge graph.
Runs Whisper on audio_file and pushes the result across modalities: the raw
audio as a shared :AssetOccurrence blob, the transcript text as a :Document,
and the Whisper segments as :TranscriptSegment nodes. Best-effort: the
ingested key is None when no engine is reachable.
CONCEPT:AU-KG.ingest.enterprise-source-extractor.
Inputs: audio_file (required), task, model, language. Returns: object
audio_transcriber_exportExportC
Export transcription to specified formats.
Inputs: result (required), formats (required).
audio_transcriber_initiate_streamInitiate streamC
Initiate the audio input stream.
Inputs: no named parameters.
audio_transcriber_interactInteractC
Interact with PersonaPlex server via WebSocket.
Inputs: server_url (required).
audio_transcriber_recordRecordC
Record audio for a specified duration or until stopped.
Inputs: seconds.
audio_transcriber_save_streamSave streamB
Save the recorded frames to a WAV file.
Inputs: no named parameters.
audio_transcriber_stop_streamStop streamA
Stop and close the audio stream.
Inputs: no named parameters.
audio_transcriber_transcribeTranscribeC
Transcribe the audio file using the initialized backend.
Inputs: fp16, task, verbose, language, +3 more.
health_checkTranscription Engine HealthARead-onlyIdempotent
Report the transcription engine's state without loading a model.
Reads which Whisper backends are importable, whether ffmpeg is on the path, the default model and the transcription directory. Changes nothing.
Inputs: no named parameters. Returns: object
transcribe_audioTranscribe AudioCIdempotent
Transcribes audio from a provided file or by recording from the microphone.
Inputs: fp16, task, model, backend, +8 more. Returns: result
transcribe_media_transcribe_segmentsTranscribe SegmentsA
Transcribe segments (transcribe_media). Verify the digest, transcribe via Whisper, and return AudioSegment-shaped
segments (start_ms/end_ms/text). Never raises — every failure
mode returns {"available": False, "error": ...}, matching the
sidecar-delegate contract's "never fabricate" invariant.
Inputs: params_json. Returns: object
transcribe_speechTranscribe SpeechARead-onlyIdempotent
Capability: speech-to-text. Transcribe one base64 audio or video payload (verified by its sha256 digest) to text with detected language, segment timestamps, optional word timestamps and a diarisation availability report. Converts any container format with ffmpeg. Returns a typed result whose status is ok or refused (refusal_code INVALID_ARGUMENT, DIGEST_MISMATCH, MEDIA_TOO_LARGE, MEDIA_TOO_LONG, UNSUPPORTED_MEDIA or TRANSCRIPTION_FAILED).
Inputs: digest (required), artifact_b64 (required), task, diarize, +3 more. Returns: object
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 12 tools
Five tools perform transcription with only subtly different inputs/behaviors (audio_transcriber_transcribe, transcribe_speech, transcribe_audio, transcribe_media_transcribe_segments, audio_ingest_transcription), making misselection likely. The stream lifecycle tools (initiate/record/save/stop) are clearer but still overlap with record-via-microphone described in transcribe_audio.
Most tools use an audio_transcriber_ prefix, but several (transcribe_speech, transcribe_audio, health_check, transcribe_media_transcribe_segments) break the convention entirely, and the latter has a redundant 'transcribe_media_transcribe_segments' pattern. The mix of prefixed and unprefixed snake_case names is inconsistent.
Twelve tools is within a reasonable range, but the count is inflated by multiple redundant transcription entry points rather than distinct capabilities. A tighter set combining transcription variants and session controls would be better scoped.
The surface covers recording, streaming lifecycle, transcription, export, health checking, and knowledge-graph ingestion, which is a broad lifecycle. Minor gaps exist around backend/model selection and stream cleanup for the PersonaPlex path, but core workflows are covered.
Maintenance
Related MCP Connectors
MCP server for Speech-to-Text
An MCP server that provides tools to discover and retrieve podcast episodes transcripts.
MCP server for RiverScript, an AI transcription platform - fetches transcripts shared via a link.
FalaAI API MCP server: transcribe audio, diagnose conversations, audit compliance (ISO 18295-1).
Related MCP Servers
- AlicenseAqualityBmaintenanceMCP server for audio transcription using local faster-whisper or OpenAI Whisper API, enabling multilingual transcription with optional GPT post-processing.3160 PyPIMIT
- AlicenseNot gradedqualityDmaintenanceMCP server for audio transcription with speaker diarization. Transcribes MP3/WAV files using Faster-Whisper and pyannote.audio, outputs markdown with speaker labels, timestamps, summaries, and action items.1MIT

Augentofficial
AlicenseBqualityDmaintenanceMCP server that turns any audio or video source into structured, searchable intelligence for agents, enabling download, transcription, semantic search, speaker identification, and more.226MIT- FlicenseNot gradedqualityDmaintenanceAn MCP server that provides speech-to-text transcription and speaker diarization using OpenAI Whisper and pyannote.audio.-