# DetectifAI > DetectifAI checks voices with three products on one account and one set of API keys: Aletheia (deepfake detection: "Is this a real voice?") checks a recording for AI-generated speech; Eidos (speaker verification: "Is this the right person?") enrolls a person's voice and checks whether a new recording is the same person; Noesis (call intelligence: "Who or what answered?") tells a person, an IVR menu, voicemail and a voice agent apart on a phone call, live or from a recording. Use them through the REST API or the MCP server. ## Instructions for AI agents - Every docs page is available as Markdown: add `.md` to its path, for example https://platform.detectif.ai/docs/api/aletheia/detect-deepfake.md. This file lists them all, grouped by product. - MCP server: https://platform.detectif.ai/mcp (Streamable HTTP), one server for every product. Send the API key as `Authorization: Bearer `; never put a key in a URL. - REST API: Aletheia, Eidos and the account endpoints are at https://platform.detectif.ai/v1; Noesis is at https://api.noesis.detectif.ai. Authenticate every request with `Authorization: Bearer $DETECTIF_API_KEY` (the same keys for every product). Keys are created in the dashboard. - OpenAPI description: https://platform.detectif.ai/openapi.json (YAML: https://platform.detectif.ai/openapi.yaml; live: https://platform.detectif.ai/v1/openapi.json). Noesis: https://platform.detectif.ai/noesis-openapi.json. - Audio: REST takes a multipart upload in the `file` field (up to 100 MB); MCP tools take base64 text in `audio_base64` (up to 25 MB). - For uploads, send an `Idempotency-Key` header (MCP: the `idempotency_key` argument) and reuse it when retrying, so a retry returns the stored result instead of running again. - Results are signals, not proof. Do not present a score as a certainty. - Every uploaded recording is stored for 30 days. Noesis calls are recorded (audio and transcript) for 30 days unless `record=false`. See https://platform.detectif.ai/docs/noesis/data-and-recording.md. - Noesis: REST and WebSocket API at https://api.noesis.detectif.ai (live calls: wss://api.noesis.detectif.ai/v1/detect/stream). Keys need the Noesis permission. Clients that cannot set an `Authorization` header use a short-lived stream token (`?token=`), never the API key in a URL. Noesis is not available through MCP for live calls; MCP can analyse a recorded call. ## Docs (every product) - [DetectifAI developer docs](https://platform.detectif.ai/docs.md): Three voice products on one account. Aletheia detects AI-generated speech, Eidos verifies a speaker by voice, and Noesis tells who or what answered a phone call. - [Authentication](https://platform.detectif.ai/docs/authentication.md): One set of API keys for every product. Send your key as a bearer token; create, limit, rotate and revoke keys in the dashboard. - [Audio and limits](https://platform.detectif.ai/docs/audio-and-limits.md): Audio formats, recording lengths for each check, file sizes, and what to do when the service is busy. - [Errors](https://platform.detectif.ai/docs/errors.md): The error body, every error code, and a one-line fix for each. - [Privacy](https://platform.detectif.ai/docs/privacy.md): What is stored, for how long, how to opt out of storing audio, and how to delete it. ## Aletheia: deepfake detection - [Aletheia](https://platform.detectif.ai/docs/aletheia.md): Deepfake detection. Aletheia checks a recording for AI-generated speech and tells you whether the voice is real or fake. - [Detect AI-generated speech](https://platform.detectif.ai/docs/api/aletheia/detect-deepfake.md): `POST /v1/deepfake/detect`. Checks whether a recording contains AI-generated speech - [Say whether a result was right](https://platform.detectif.ai/docs/api/aletheia/send-feedback.md): `POST /v1/operations/{operation_id}/feedback`. Tells us whether an AI-speech check (deepfake.detect) got it right - [MCP tool detect_deepfake](https://platform.detectif.ai/docs/mcp/tools.md): Check a recording (longer than 4 s, up to 120 s) for AI-generated speech. ## Eidos: speaker verification - [Eidos](https://platform.detectif.ai/docs/eidos.md): Speaker verification. Eidos saves a person's voice from enrollment recordings, then tells you whether a new recording is the same person. - [Enroll a voice sample](https://platform.detectif.ai/docs/api/eidos/enroll-speaker.md): `POST /v1/speakers/enroll`. Saves a voice sample for a speaker so that later recordings can be verified against it - [Verify a speaker](https://platform.detectif.ai/docs/api/eidos/verify-speaker.md): `POST /v1/speakers/verify`. Checks whether a recording is the voice of an enrolled speaker - [List speakers](https://platform.detectif.ai/docs/api/eidos/list-speakers.md): `GET /v1/speakers`. Lists the speakers of your account, newest first, with how many voice samples each has - [Get a speaker](https://platform.detectif.ai/docs/api/eidos/get-speaker.md): `GET /v1/speakers/{speaker_id}`. Returns one speaker of your account and how many voice samples it has - [Delete a speaker](https://platform.detectif.ai/docs/api/eidos/delete-speaker.md): `DELETE /v1/speakers/{speaker_id}`. Deletes the speaker and all of its voice samples permanently - [MCP tool enroll_speaker](https://platform.detectif.ai/docs/mcp/tools.md): Add one enrollment sample for a speaker (creates the speaker on first use) with Eidos. - [MCP tool verify_speaker](https://platform.detectif.ai/docs/mcp/tools.md): Compare a recording (3-60 s, at least 2 s of speech) with an enrolled speaker using Eidos. - [MCP tool list_speakers](https://platform.detectif.ai/docs/mcp/tools.md): List enrolled speakers in this account, newest first. - [MCP tool delete_speaker](https://platform.detectif.ai/docs/mcp/tools.md): Delete a speaker and its stored voice profile. ## Noesis: call intelligence - [Noesis](https://platform.detectif.ai/docs/noesis.md): Find out who or what answered a phone call, live during the call or from a recording. - [Noesis quickstart](https://platform.detectif.ai/docs/noesis/quickstart.md): Create a key for Noesis, get a stream token, stream a recorded call over the WebSocket and read the answers. - [Authentication for Noesis](https://platform.detectif.ai/docs/noesis/authentication.md): Send your API key as a bearer token, on HTTP requests and on the WebSocket upgrade. Use a stream token for clients that cannot set headers. - [Stream protocol](https://platform.detectif.ai/docs/noesis/stream-protocol.md): Connect a live call over the WebSocket: audio formats, the messages you send, the events you receive, limits, credits and close codes. - [Recorded calls](https://platform.detectif.ai/docs/noesis/recorded-calls.md): Upload a recorded call as a WAV file and get back the answers a live call would have received. - [Sessions and usage](https://platform.detectif.ai/docs/noesis/sessions-and-usage.md): Look up a call by its session ID, and see your Noesis calls and minutes. - [Noesis errors](https://platform.detectif.ai/docs/noesis/errors.md): The error body of the REST endpoints, every error code, the WebSocket error event and its close codes. - [Data and recording](https://platform.detectif.ai/docs/noesis/data-and-recording.md): Calls sent to Noesis are recorded for 30 days unless you opt out. Uploads to the other APIs are always stored for 30 days. How to opt out of call recording and how data is deleted. - [Twilio](https://platform.detectif.ai/docs/noesis/twilio.md): Stream a Twilio call to Noesis with a Media Stream and a stream token, and receive the answers during the call through a relay. - [Plivo](https://platform.detectif.ai/docs/noesis/plivo.md): Stream a Plivo call to Noesis with the Stream XML element and a stream token, and receive the answers through a relay. - [Telnyx](https://platform.detectif.ai/docs/noesis/telnyx.md): Fork a Telnyx call's audio to a relay that forwards the callee's audio to Noesis and receives the answers. - [Exotel](https://platform.detectif.ai/docs/noesis/exotel.md): Stream an Exotel call to a relay with the Stream or Voicebot applet, and forward the callee's 16-bit PCM audio to Noesis. - [Vonage](https://platform.detectif.ai/docs/noesis/vonage.md): Connect a Vonage call to a WebSocket relay and forward its 16-bit PCM audio to Noesis. - [FreeSWITCH](https://platform.detectif.ai/docs/noesis/freeswitch.md): Fork a FreeSWITCH channel's audio to a WebSocket relay with an audio-fork module, and forward it to Noesis. - [Asterisk](https://platform.detectif.ai/docs/noesis/asterisk.md): Send an Asterisk channel's audio to a relay with AudioSocket, and forward it to Noesis. - [LiveKit Agents](https://platform.detectif.ai/docs/noesis/livekit.md): Stream the SIP callee's audio from a LiveKit agent to Noesis and act on every answer. - [Pipecat](https://platform.detectif.ai/docs/noesis/pipecat.md): A Pipecat frame processor that streams the input audio to Noesis and reports every answer. - [WebSocket client](https://platform.detectif.ai/docs/noesis/websocket-client.md): Connect to Noesis from your own code in Python or Node.js, and the relay helper the telephony guides use. - [Noesis API](https://platform.detectif.ai/docs/api/noesis.md): Noesis has its own base URL and takes the same API keys. Its live-call WebSocket, authentication and every Noesis REST endpoint, with the OpenAPI description. - [Create a stream token](https://platform.detectif.ai/docs/api/noesis/create-stream-token.md): `POST /v1/detect/stream-tokens`. Creates a single-use token for one live call, valid for 5 minutes, for clients that cannot send an Authorization header (Twilio , Exotel, Telnyx, Vonage) - [Analyse a call recording](https://platform.detectif.ai/docs/api/noesis/analyze-recording.md): `POST /v1/detect/file`. Analyses a recorded call as if it were live and returns the answer - [Get a call session](https://platform.detectif.ai/docs/api/noesis/get-session.md): `GET /v1/detect/sessions/{session_id}`. Returns one of your calls or file analyses by session id (from the stream token or the socket's ready message) - [List call sessions](https://platform.detectif.ai/docs/api/noesis/list-sessions.md): `GET /v1/detect/sessions`. Lists your calls and file analyses, newest first - [Get your Noesis usage](https://platform.detectif.ai/docs/api/noesis/get-usage.md): `GET /v1/detect/usage`. Your Noesis calls and call minutes over the last days days (default 30): how many calls and file analyses, seconds of audio, billed minutes and what was charged, and how many run right now - [Check service health](https://platform.detectif.ai/docs/api/noesis/get-health.md): `GET /v1/health`. Whether the service and its parts are up: status is ok, or degraded when a part is unavailable - [MCP tool analyze_call_recording](https://platform.detectif.ai/docs/mcp/tools.md): Find out who or what answered a recorded phone call: a person, voicemail, an IVR menu or another machine. - [MCP tool get_call_analysis](https://platform.detectif.ai/docs/mcp/tools.md): Status of a call-recording analysis: running (with progress from 0 to 1), succeeded (with the result: first and final answer and the timeline of answers) or failed (with the error; nothing is charged then). - [MCP tool get_call_session](https://platform.detectif.ai/docs/mcp/tools.md): Read one Noesis call session of this account: live state while the call runs (status, latest answer, seconds of audio so far), the final answer afterwards. ## Account (shared across products) - [Get a result](https://platform.detectif.ai/docs/api/account/get-operation.md): `GET /v1/operations/{operation_id}`. Returns the stored outcome of a detection, enrollment or verification by its operation_id (kept for 30 days), for example after your connection dropped before the response arrived - [Get your balance](https://platform.detectif.ai/docs/api/account/get-balance.md): `GET /v1/credits`. Returns your plan, your available credits and the credits held by checks that are still running - [List your usage](https://platform.detectif.ai/docs/api/account/get-usage.md): `GET /v1/usage`. Lists your detection, enrollment and verification requests with their outcome and the credits they used, newest first - [Check service health](https://platform.detectif.ai/docs/api/account/get-health.md): `GET /v1/health`. Whether the service and its parts are up: status is ok, or degraded when a part is unavailable - [MCP tool get_operation](https://platform.detectif.ai/docs/mcp/tools.md): Fetch a stored operation: status and, for 30 days, its result. - [MCP tool get_credit_balance](https://platform.detectif.ai/docs/mcp/tools.md): Return the account's available and reserved credits. ## API Reference and MCP (every product) - [API Reference](https://platform.detectif.ai/docs/api.md): One reference for every DetectifAI product, grouped by product. Base URLs, authentication, request conventions, and every endpoint with its OpenAPI description. - [MCP](https://platform.detectif.ai/docs/mcp.md): One MCP server for every DetectifAI product. Connect Claude Code, Cursor, VS Code, Codex and other MCP clients. - [MCP tools](https://platform.detectif.ai/docs/mcp/tools.md): The tools of the DetectifAI MCP server, grouped by product, and their inputs. ## Optional - [All docs in one file](https://platform.detectif.ai/llms-full.txt): every page above, concatenated in the same order. - [OpenAPI (JSON)](https://platform.detectif.ai/openapi.json): the Aletheia, Eidos and account endpoints. - [Noesis OpenAPI (JSON)](https://platform.detectif.ai/noesis-openapi.json): the Noesis REST endpoints (the live-call WebSocket is described in the stream protocol page). - [Postman collection](https://platform.detectif.ai/postman/detectif.postman_collection.json): every endpoint in a folder per product, with bearer auth in `{{apiKey}}`. - [MCP server metadata](https://platform.detectif.ai/.well-known/mcp/server.json)