TalkToHermes
AF360/TalkToHermes
Native iPhone and iPad push-to-talk voice client for Hermes Agent with self-hosted STT and TTS
TalkToHermes is a native SwiftUI voice client for iPhone and iPad that lets you speak to your own Hermes Agent with push-to-talk, using speech recognition and synthesis you host yourself.
What TalkToHermes does
TalkToHermes combines an iOS app with a per-user TalkToHermes Voice Bridge that connects the phone to a Hermes Agent. You hold to talk, follow the full conversation on screen, and handle live approval and cancellation flows. Credentials are kept in the Keychain, the interface is available in English and German, and app language and spoken language are chosen independently.
Speech recognition and synthesis run through configurable fallback chains. For recognition the README names Faster-Whisper on a GPU system, Wyoming Faster-Whisper, MLX-Whisper on an Apple Silicon Mac, or a local Hermes STT model. For synthesis it names local Piper as the last resort, Wyoming-Piper, and optional OmniVoice for cloned voices. Audio files stay on your infrastructure even when Hermes uses a cloud LLM, and only transcribed text and conversation context go to the model provider. With a local model provider such as Ollama the whole interaction stays local.
Key features
- Native SwiftUI push-to-talk client with full conversation history
- Ordered fallback chains for speech-to-text and text-to-speech providers
- Live approval and cancellation flows
- Keychain-backed credentials
- English and German interface with separate spoken-language setting
- Optional cloned voices through OmniVoice
When to use it
- Talking to a home-hosted Hermes agent from an iPhone
- Keeping voice audio on your own network while using a cloud LLM
- Approving or cancelling an agent action by voice session
Who it is for: iPhone and iPad users who run their own Hermes Agent and want a private voice interface.
How it fits with Hermes Agent
Built for Hermes Agent: the app talks to a Hermes instance through the TalkToHermes Voice Bridge.
Requirements: A configured Hermes Agent, the TalkToHermes Voice Bridge, at least one STT and one TTS provider, HTTPS access to the bridge, an iPhone or iPad on iOS 17+, and Xcode if building from source
Note: It is not standalone: it needs a configured Hermes Agent, the Voice Bridge, speech services and a private HTTPS endpoint.
FAQ
What is TalkToHermes?
TalkToHermes is a private, native iPhone and iPad voice client for Hermes Agent. It offers push-to-talk, a full on-screen conversation and self-hosted speech recognition and synthesis.
Does TalkToHermes work with Hermes Agent?
Yes. It is built for Hermes Agent and connects to it through a per-user TalkToHermes Voice Bridge.
What do I need to run TalkToHermes?
You need a configured Hermes Agent, the Voice Bridge, speech-to-text and text-to-speech providers, and HTTPS access to the bridge. On the device you need iOS 17 or later, plus Xcode to build from source or an unsigned IPA from the latest release for sideloading.
Similar interfaces for Hermes Agent
All interfacesAlways-on 'Okay Hermes' wake-phrase voice daemon for Hermes Agent on Linux
skalenetwork ClawbitsTeam chat where AI agents are members with their own mailbox, git repos and automations
1941s Hermes-Agent UILocal-first web UI for Hermes Agent with streaming reasoning traces, artifacts and Docker Compose deploys
TNJ2026 Hermes DeckNative macOS SwiftUI client that chats with Hermes, Hermes profiles and external coding agents
styles01 Stack-chan x OpenClaw x HermesESP-IDF firmware and bridge that make an M5Stack Stack-chan robot a node of OpenClaw or Hermes agents
zd87pl Hermes G2 HUDMonitor and control Hermes agents from Even Realities G2 smart glasses through a WebSocket gateway
Related guides: Connect Hermes agents on several machines with Hermes Desktop · How to install Hermes Agent