Hermes Speech

What Hermes Speech does

Hermes Speech registers allmodels as a normal Hermes TTS and transcription provider, so Hermes' existing /voice mode, sentence pipeline, messaging gateways, CLI, TUI and desktop app keep working. Spoken replies and voice-message transcription go through Hermes' standard provider interfaces. Models that support streaming TTS use it; other models keep Hermes' synchronous sentence pipeline.

Setup is conversational: tell the agent to set up AllModels speech and it handles signup, email verification and starting defaults, which are fish/s2-1-pro for TTS and soniox/stt-async-v5 for transcription. Voices can be searched by name, description, language, gender, provider or quality and previewed before selection. Later requests can change models and voices, tune speed, check balance, create top-up links, test audio and check for plugin updates. The plugin uses the OpenAI and HTTP clients bundled with Hermes, so it needs no extra pip install.

Key features

• Registers AllModels as a Hermes text-to-speech and speech-to-text provider
• Conversational setup that handles signup and default voice and transcription models
• Voice search by name, description, language, gender, provider or quality, with audio previews
• Streaming TTS where the model supports it, with Hermes' sentence pipeline as the fallback
• Speech management to change models and voices, tune speed, check balance and create top-up links
• No additional pip install; it uses the openai and httpx libraries bundled with Hermes

When to use it

• Giving a Hermes gateway bot spoken replies in a chosen voice
• Transcribing voice messages sent to Hermes
• Switching voices or speech models by asking the agent in chat

Who it is for: Hermes Agent users who want TTS and STT from a single provider account and prefer to configure it by talking to the agent.

How it fits with Hermes Agent

Built for Hermes Agent as a plugin that registers allmodels as a standard TTS and transcription provider.

How to install Hermes Speech

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

hermes plugins install allmodels-io/hermes-speech --enable

Requirements: Hermes Agent 0.20.0 or later and an AllModels account; no additional Python packages.

Note: Speech runs through the AllModels service, which needs an AllModels account and a balance that can be topped up.

FAQ