Ichigo: Local real-time voice AI
21–30 of 45 posts
Re: Ichigo: Local real-time voice AI
#22Re: Ichigo: Local real-time voice AI
#23This is a really cool project! What have people built with it? I'd love to learn about what local apps people are building on this.
Thanks! We've received feedback on use cases like live translation, safe and untrackable educational tools for kids, and language-learning apps. There are so many possibilities, and hope to see guys building amazing products on top of Ichigo.
I'm trying to use chatgpt for ai translation, but the other big problem I run into is TTS and SST on non-top 40 languages (e.g. lao). Facebook has a TTS library, but it isn't open for commercial use unfortunately.
Re: Ichigo: Local real-time voice AI
#24Earlier quoted context omitted.
There are strawberries all over the readme so I reck you're right.
Is this a continuation of the meme that GPT can't identify the number of "R"s in "strawberry"?
GPT 4o: The word "ichigo," which is the Romanized spelling (romaji) of いちご, contains one "r." It appears in the letter "r" in "chi," as the "ch" sound in romaji represents a combination of the "r" sound from "r" and "t" sound from "i."
Thank you chatgpt. I'm glad we've burned down a bunch of forests for this.
You can consistently get the right answer with a prompt of:
> Write python code, and run it, to count the number of 'r' characters in いちご.
though. For numeric stuff, telling the thing to just write python code makes it significantly better at getting right answers.
Re: Ichigo: Local real-time voice AI
#25Very cool, but a bit less practical than some alternatives because it does not seem to do request transcription.
Actually, it does. You can turn on the transcription feature from the bottom right corner and even type to Ichigo if you want. We didn’t show it in the launch video since we were focusing on the verbal interaction side of things.
To clarify, while you can enable transcription to see what Ichigo says, Ichigo's design skips directly from audio to speech representations without creating a text transcription of the user’s input. This makes interactions faster but does mean that the user's spoken input isn't transcribed to text.
The flow we use is Speech → Encoder → Speech Representations → LLM → Text → TTS. By skipping the text step, we're able to speed things up and focus on the verbal experience.
Hope this clears things up!
Re: Ichigo: Local real-time voice AI
#26Great initiative! before adding more comments, I'm trying to set up on my local Mac M3 machine. I'm having a hard time to install dependencies. Anyone here have the same issue?
Re: Ichigo: Local real-time voice AI
#27Earlier quoted context omitted.
Thanks! We've received feedback on use cases like live translation, safe and untrackable educational tools for kids, and language-learning apps. There are so many possibilities, and hope to see guys building amazing products on top of Ichigo.
I just tried to use the demo website for live translation. The AI always responded in English, either ignoring my request to only respond in French or Lao, or preface the translation with english ("I can translate that to French. the translation is: ..."). I'm trying to use chatgpt for ai translation, but the other big problem I run into is TTS and SST on non-top 40 languages (e.g. lao). Facebook has a TTS library, b…
Re: Ichigo: Local real-time voice AI
#28Finally I can use one of the random facts that have entered my brain for decades now even though I can't remember where my keys are. If I remember correctly, "ichigo" means strawberry in japanese. You are welcome.
Re: Ichigo: Local real-time voice AI
#29Bringing AI into this space enhances user experience while respecting their autonomy over data. It feels like a promising step toward a future where we can leverage the power of AI without compromising on privacy or control. Really looking forward to seeing how this evolves!
Re: Ichigo: Local real-time voice AI
#30Emre here from Homebrew Research. It's great to see Ichigo on HN! A quick intro: We're a Local AI company building local AI tools and training open-source models. Ichigo is our training method that enables LLMs to understand human speech and talk back with low latency - thanks to FishSpeech integration. It is open data, open weights, and weight initialized with Llama 3.1, extending its reasoning ability. Plus, we are…