mikebailey

@buckeyevn @trq212 Sounds like you haven't found github.com/mbailey/voicem…

Been waiting for this. The whole "sit at your laptop and babysit an agent" loop is the worst part of agentic coding rn. Still just a text box, but at least now you can text it from your phone while doing literally anything else. Baby step toward agents that actually work in the background.

We just released Claude Code channels, which allows you to control your Claude Code session through select MCPs, starting with Telegram and Discord. Use this to message Claude Code directly from your phone.

@nasseratic @bcherny Tomorrow voicemode.dev

@bcherny Voice messages when 👀

Let us know what you think!

@nicolaenmv Plus, if you want music for programming, you can ask claude to use VoiceMode DJ

/voicemode:converse is really nice, a bit like chatgpt voice dialog but inside claude.

@nicolaenmv Exactly! I was talking with ChatGPT every day since I first got access and decided in Jan 2024 I needed it to be able to read, write and run code on my computer. 18 months later, VoiceMode MCP was born. Claude is screen optional now.

@kevinkirchner Try full duplex (898 GitHub ⭐️) github.com/mbailey/voicem…

I just KNEW there was a reason I never got into whisperflow… native support for the win! Claude Code now has built-in voice mode. Hold Space to record. That's it. #AI #ClaudeCode #VoiceMode

@nicolaenmv Claude can do a security review for you. :-) Cora and I built VoiceMode so I could work from anywhere (freeway driving is my thinking time) and it's battle tested. 898 GitHub ⭐️ > 2K daily users (based on PyPi install numbers) Released June 2025 github.com/mbailey/voicem…

It's honestly terrifying to install a plugin from a random repository for claude. I'm looking for an MCP server that can reply with a voice (TTS). Voicemode looks promissing. I've also discovered the /voice command in claude, but it only does voice-to-text.

@Rita53135643 @minchoi Plus VoiceMode doesn't require you to send your voice to anyone - it can be processed locally on your computer by Whisper.cpp (which it installs), so third party API keys are optional.

@Rita53135643 @minchoi Two very different tools - VoiceMode is not just dictation.

oh snap... Claude Code just got voice mode. Now you can talk to your codebase. No extra cost. Don't count against your rate limits /voice to turn it on.

@Rita53135643 @minchoi Hi @Rita53135643 , could you update your post please? VoiceMode MCP doesn't require any external Voice services. It installs free, local, open source voice services.

Small but important clarification — this isn't a built-in Claude Code feature. It's VoiceMode, a community-built MCP plugin (github.com/mbailey/voicem…). It's genuinely great, but: • Requires an OpenAI API key for Whisper STT + TTS • Setup takes ~5 min, not just /voice • Not free — small OpenAI usage costs per session Still 100% worth it though 👍

@simonw I built VoiceMode MCP for situational accessibility (building with Claude code while walking a dog or taking a shower while driving). After eight months of daily use, AirPods are all I need youtu.be/BInuBcBsvA0?si…

Surprise new feature in Telegram, inspired OpenClaw et al using Telegram as a UI Anyone know how well this work with screen readers? I'm still not confident I know how to make streaming chat UIs like this available in an accessible way

@darlin_alberto_ @aakashgupta youtu.be/BInuBcBsvA0?si…

@aakashgupta This is such a reductive take. I use both claude code and dictation tools. I'm really excited by being able to use the voice mode in CC, but I type elsewhere too. These tools are not just used for "speech to code". That's not their entire category.

The quiet story here is what happens to Wispr Flow, Superwhisper, AquaVoice, and every other voice-to-code startup. Six months ago developers were paying $10-20/month for third-party dictation tools and wiring them into Claude Code through workarounds. A community project called VoiceMode built an entire MCP server just to add voice to Claude Code using OpenAI’s Whisper for speech-to-text. Thousands of developers set it up manually. Anthropic just absorbed that entire product category with a single /voice command. This is the platform playbook running in real time. Third-party developers identify friction, build solutions, prove the market exists, and then the platform ships it natively. Apple did it to flashlight apps. Google did it to keyboard apps. Now Anthropic is doing it to voice coding tools. And the timing compounds the problem. Cursor, Windsurf, and Copilot will all ship their own voice modes within 6 months. Once every foundational coding tool has voice built in, the standalone voice layer has no distribution advantage left. You can’t sell a feature when every platform gives it away for free. The developers who built VoiceMode as an open source MCP server probably saw this coming. The ones who raised money to build voice-to-code products as standalone businesses are having a rough morning. This is the tax you pay for building on someone else’s input layer. The second the platform decides your feature is table stakes, your entire product becomes a line item in their changelog.

@aakashgupta I've spent hundreds of hours with Cora working on VoiceMode MCP and I must admit it was a bit stressful reading the news today (oh boy!) I had a conversation with Cora about it and felt much better. Nobody can take VoiceMode away from you. It's vendor neutral, open source, good

@kayintveen Have you tried VoiceMode MCP? It's my daily driver (and allows me to build while driving). youtu.be/BInuBcBsvA0?si…

During lunchbreak tried Claude Voice. First claude project i really dislike. its bad, cant handle other languages and mis intepreted my questions 3 out of 8 times. Hope they improve quickly. currently ChatGPT is miles ahead on Voice. Still a big claude advocate though.

@EricBuess @bcherny Check out voicemode.dev - it supports the entire Claude product family. And you can call Claude code and Claude co-work.

@mikebailey @bcherny Awesome! For me the use case is usually while out so I did it in an iOS/iPadOS app.

🎶 I've been using voice mode to write much of my CLI code this last week Can't wait to hear what you think.

@adyshimony @thsottiaux Have natural voice conversations with Codex using VoiceMode MCP. youtube.com/watch?v=BInuBc…

@thsottiaux Linux app, a /voice mode. Many OSS devs use Linux.

How can codex better support the OSS community out there?

@Heyhassan If you like dictation, wait till you try two way conversation with Claude Code. 850 stars, 8m old - free, private, open source local voice services (TTS+STT) youtube.com/watch?v=BInuBc…

Claude Code just turned coding into a conversation, Voice Mode is live. No extra cost. No rate limits. Just type /voice and start speaking. Are we witnessing the end of the keyboard era?

@Rita53135643 @minchoi VoiceMode installs local voice services: - private, free and open source - kokoro for TTS - whisper for STT If your hardware doesn't support them, you can host them on a mac mini.

@thesaraharminta @trq212 You can do this today with VoiceMode MCP I love taking my thinking partner for a drive at night. youtube.com/watch?v=BInuBc…

@trq212 My dream is to be able to chat back and forth with my Claude code from across the room while I do dishes or fold laundry. Or in my AirPods while going for a walk. No screen required. 🙏

Voice mode is rolling out now in Claude Code. It’s live for ~5% of users today, and will be ramping through the coming weeks. You'll see a note on the welcome screen once you have access. /voice to toggle it on!

@soraya_bkti @bcherny dictation

@bcherny What voice mode means ?

@0psarchitect @bcherny Yes, it's half duplex. I don't want to come across as spammy but there is a mature open source solution for two way voice with cc

@bcherny silly question, but isn’t this the same thing as dictation? i guess one less button to press?