Ermine.ai is an AI tool that enables users to transcribe audio directly from their device microphone, using 100% local/client-side processing.

This means that the transcription is performed using the user's own device, without the need for any external servers or internet connection.

The tool is available for download from GitHub, and offers users the option to download both the audio file and transcript for later use.

However, before the transcription process can begin, the tool requires the user's browser to load and initialize the transcription model.

This may take a few minutes during the first use while the model files (approximately 50mb) are downloaded and cached.

The model currently only supports English transcription, and the tool may prompt users to allow microphone access in order to initiate the transcription process.

Ermine.ai offers an efficient and secure way to transcribe audio recordings, especially for those who are concerned about privacy and data security.

By choosing to use client-side processing, the user's audio data remains within their own device and doesn't travel to external servers or the cloud.

Additionally, the tool's ability to download both the audio and transcript for future use enhances its usability and ensures that users can access their transcriptions at their convenience.