Hermes Voice Transcript: messaging bridge for Hermes Agent

Hermes plugin that stops the agent from hunting for audio files it has already transcribed

What Hermes Voice Transcript does

Hermes Voice Transcript fixes one specific behavior. When a voice note arrives, Hermes transcribes it, but the upload filename travels in the same message as the transcript. The agent answers correctly and then goes looking for the audio file it has already heard. The plugin drops the filename at the one place where Hermes joins transcript and caption, and leaves the transcript in place.

It wraps a single method, GatewayRunner._enrich_message_with_transcription, in shared gateway code, so voice notes on Matrix, Telegram and Signal get the same treatment. A caption is removed only when its extension is one the transcriber can handle, so a real caption you typed with the recording survives. There is nothing to configure and no key. A log line on load shows what was wrapped, and grep -a voice-transcript on the agent log finds it. The plugin needs a working STT provider, Python 3.11 or newer, and no third-party dependencies.

Key features

Removes the audio filename from the transcript message
Works in shared gateway code for Matrix, Telegram and Signal voice notes
Leaves real captions untouched when the extension is not transcribable
No configuration, key or third-party dependency
Load-time log line plus warnings if the target method changes

When to use it

Stopping a Hermes agent from running a dozen tool calls to find a voice note it already transcribed
Using voice notes with Hermes on Matrix, Telegram or Signal without wasted tool use

Who it is for: Hermes Agent users who talk to their agent through voice notes in a messaging gateway.

How it fits with Hermes Agent

Built for Hermes Agent as a plugin that wraps one internal gateway method. The README says it was tested against Hermes Agent v0.20.0.

How to install Hermes Voice Transcript

These commands are copied from the project's README. Check the repository for the latest steps before you run them.

hermes plugins install etkecc/hermes-voice-transcript --enable

Requirements: Hermes Agent with a working STT provider, and Python 3.11 or newer

Note: It wraps the internal method GatewayRunner._enrich_message_with_transcription, so an upstream rename or move would disable it quietly.

FAQ

What is Hermes Voice Transcript?

Hermes Voice Transcript is a Hermes Agent plugin that stops the agent from searching for a voice note's audio file after Hermes has already transcribed it. It removes the filename from the message and keeps the transcript.

How do I install Hermes Voice Transcript?

Run hermes plugins install etkecc/hermes-voice-transcript --enable and restart the gateway. There is nothing to configure.

How do I know Hermes Voice Transcript is working?

Send a voice note. If the agent answers what you said without reaching for tools, it works. You can also run grep -a voice-transcript on the Hermes agent.log to see the load-time line.