> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://murf.ai/api/docs/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://murf.ai/api/docs/_mcp/server.

## Finch (Beta) on AWS Marketplace

You can run Finch speech to text (Beta) inside your own AWS account. Subscribe on AWS Marketplace, deploy a SageMaker endpoint, and call it directly. There is no Murf API key, and audio and transcripts stay in your account. AWS bills the usage.

Finch Global covers English and six European languages. Finch Indic covers 11 Indian languages and English (India), and returns word timestamps. Transcribe one file, stream live audio, or run a batch of files from S3.

The listing is on [AWS Marketplace](https://aws.amazon.com/marketplace/pp/prodview-jjx4ovbqssnpq).

Read the docs

## Consent verification for voice cloning

Creating a voice clone now requires a short recording of the speaker reading this statement: "This is my own voice, and I consent to Murf using this recording to create an AI clone of my voice." Send it as `consentAudio` with a file upload, or as `consentAudioUrl` when you pass a public audio URL. Murf checks that the recording matches the statement and that it is the same person as the voice sample. If the check fails, the creation status comes back as `FAILED`.

Voice cloning is available on Free, Pay as you go, and Enterprise. A workspace can keep 1 clone on Free, 3 on Pay as you go, and 5 on Enterprise.

Read the docs

## Zero data retention for Speech to Text

Enterprise plans can set `zero_retention=true` on the Speech to Text streaming WebSocket. Streaming audio and transcripts are not stored. File upload and bulk transcription keep the transcript for 24 hours so you can download it.

Read the docs