Show HN: TalkiTo – enabling voice and Slack for Claude Code and Codex CLI

github.com

5 points by robbomacrae a day ago

Hey everyone, here is an open source project I've been working on to add voice input/output to terminal based coding agents.

One thing about the new terminal coding agents I really like is being able to multi-task but right now it's a bit like a Tesla on autopilot needing your hands still on the wheel. You need to be checking often if your input is required or if it's going off the rails. To be able to go fully hands free I wanted to add TTS and ASR. Then I added slack and WhatsApp hooks to TalkiTo as well.

It's fully open source with a BYOK philosophy and it's configured to work with any of the major ASR/TTS providers. It also supports local whisper and kokoro/kittentts if you want a decent free/private option.

It works by wrapping the coding agent and capturing the input/output. It does have an MCP server running but thats mainly for configuration - I found that using MCP to speak or listen was too slow. The upshot of the MCP server is you can type (or say) "talkito disable ASR" or "talkito change tts to kokoro".

Here is a demo video I made here: https://www.youtube.com/watch?v=pf8jFt0smqs

I like to think of it as similar to SuperWhisper but with TTS, the focus on coding agents and configurability.

Really curious to get feedback.

Thanks!

yodon a day ago

This is likely a separate tool, but I suspect it is trivial to build given the framework you've set up...

What I most want is something that notices if Claude has been thinking for more than say 30 seconds, and if so sends me a slack message when the output is finished and ready to read.

Given that Claude is often thinking for 30 minutes, I context switch onto something else, and wish there were a way to get a notification so it isn't sitting idle waiting for me for ten minutes (or an hour, as sometimes happens)

  • robbomacrae a day ago

    You've hit the nail on the head of one of the main motivators of this project! I always context switch and miss when its done its task or sometimes forget completely..

    So essentially at present when you setup TalkiTo with a Slack bot (instructions are shown by running `talkito --setup-slack`) it will do this its just its sending all the output to slack not just the final completion. I have verbosity levels and I could try tweaking them so that -verbosity 0 essentially does what you want ie filter everything except for prompts to the user. If you put in an issue on GitHub I'll get to it!

    • yodon a day ago

      Would you consider a lightweight app that just does notification?

      Installing and running Whisper locally just to get notifications seems like more resource consumption than some of us are looking for.

      Some will want one, some will want both.

      • robbomacrae a day ago

        Yeah that sounds reasonable. I've been thinking about how to make it more modular. I'll make it so that base TalkiTo doesn't need any heavy libraries.

        EDIT: also to be clear whisper doesn't run or even download the model unless it's selected and the user confirms the DL. This can be configured. Google cloud has a free ASR option and ASR can be switched off completely.