Harry Coultas Blum (@harrycblum)
Fluxions AI가 음성 기반 바이브 코딩 도구 vui를 출시했다는 소식입니다. 스레드 작성자는 자신들도 같은 방향에 도달했다며, StepFun 기준의 첫 토큰 지연시간(first-token latency)이 얼마나 나오는지 질문하고 있습니다. 음성 인터페이스와 저지연 LLM 추론이 중요한 실사용 영역에 해당합니다.
Harry Coultas Blum (@harrycblum)
Fluxions AI가 음성 기반 바이브 코딩 도구 vui를 출시했다는 소식입니다. 스레드 작성자는 자신들도 같은 방향에 도달했다며, StepFun 기준의 첫 토큰 지연시간(first-token latency)이 얼마나 나오는지 질문하고 있습니다. 음성 인터페이스와 저지연 LLM 추론이 중요한 실사용 영역에 해당합니다.
jason (@jasonth0)
작성자는 코드 작성에서 음성 입력(voice input)이 '치트 코드' 수준으로 개발 속도를 비약적으로 올려주고 있다고 평가합니다. 음성 기반 코딩(workflow)에 대한 개인적 경험 공유로, 음성-코드 인터페이스가 개발 생산성 향상에 유용하다는 실사용 사례입니다.
Omnara: Ứng dụng giúp lập trình bằng giọng nói, không cần bàn phím! 🤯 Demo tạo game Pong từ đầu, chỉnh sửa giao diện cyberpunk chỉ bằng lệnh thoại. Dùng Claude Code SDK, cho phép chạy các tác vụ song song. Thử nghiệm ngay!
#laptrinh #AI #tool #voicecoding #Omnara #trituenhantao #congnghe
🚀 Meet Grok Chat by You.com – a new AI assistant that lets you debug, build, and test web apps using voice and images. Think ChatGPT meets a visual IDE, with open‑source vibes and real‑time code suggestions. Curious how it works? Dive into the details and see it in action! #GrokChat #AIdebugging #VoiceCoding #WebAppAI
🔗 https://aidailypost.com/news/grok-chat-ai-debugging-building-testing-web-apps-voice-images
Vừa xuất bản một bài viết mới bàn về hướng đi của 'vibe coding', 'voice coding' và lập trình có sự hỗ trợ của AI. Chủ đề này mở ra nhiều tiềm năng trong tương lai của ngành lập trình.
#VoiceCoding #AICoding #FutureOfProgramming #LậpTrìnhGiọngNói #LậpTrìnhAI #CôngNghệ
"Vide coding" is a hot topic right now, with new AI tools allowing you to describe what you want to work on with a prompt, and it appears in front of your eyes.
I look at Aqua Voice, which lets you "vibe code", take notes, write, and much more and as you can see from Premiere's transcript, it handles things better!
https://www.youtube.com/watch?v=uAv39wdf_0k
#ai #vibecoding #voicecoding #voiceassistant #dictation #aiassistant
Day 6
So far the big challenge has been improving accuracy, I started out with the microphone from my headset, it's a Beyerdynamic, a fancy audio brand, so I expected it to be ok, but it actually wasn't that great.
I then switch to a condensor mic I've had for some twenty years, but it seems that one has given the ghost. It works but produces a lot of rumble. Not that surprising, these kind of microphones are notably fragile.
Now I'm using a Shure SM58. It's a classic magnetic coil mic. Indestructible and well suited for vocals. Now it's much better. I also did some vocal warmup this morning and lo and behold that helped too!
There are still a few things it regularly gets wrong, probably cause I'm not enunciating things well enough. E.g. I say "tilde" and it types "the", or "brace" instead of "phrase". Making an effort to put on my "radio host" voice.
Now that accuracy is much better I can more confidently start chaining longer commands, which is essential to get fast enough. And I'm continuing to add custom commands for stuff I do regularly, so I don't have to "spell" things using the talon alphabet. E.g. "change directory" for "cd ", instead of "cap drum space". This also helps with accuracy (more recognizable phrases), and provides easier mnemonics.
I've made a recording of myself now that I'm a few days in using talon voice, as a snapshot to show where I am in the process.
You can see me use emacs and firefox using my voice, while I try to get something working in clojurescript.
It helps that I'm using vim style bindings with emacs, all eventually learn or set up custom editor commands, but for now as long as I can spell letters I'm good to go.
I’m slowly getting better at #voicecoding
If you haven’t checked out Talon Voice and you experience #rsi then I cannot recommend it enough.