live speech and tts
in progress
Hans Henrik Bergan
Merlin is being lazy as fuck. TTS is built-in to modern web browsers under the HTML5 technology window.speechSynthesis + window.SpeechRecognition , it would cost Merlin 1-3 days of engineering effort to implement, and wouldn't cost them anything in back-end infrastructure. It's been requested for years, in multiple feature-requests. I implemented this in my own LLM gateway in a single day, and it works great in Google Chrome, both Desktop and Android, meaning the vast majority of users. (Then again, I'm a senior engineer with ~20 years experience. I would expect a junior engineer to take more time)
A
Andrew Zaragoza
ETA on this? I'd love to see this come to fruition.
b
brunocosta77
How is this one Vijay Bharadwaj?
Y
Yume
pfft, another release a feature then limit its capacity, is not a win here. Then you will pay more for credit output, that has daily limits. Nah. We're good. More features. More limits...
Vijay Bharadwaj
updated the status to
in progress
I have set this in motion. Product + design + proof of concepts will take a week or two. Development follows that.
Sven Kernke
Does that mean it involves two parts? One “input” via voice and an output as language, or just one of the two? If yes which? :) endu

endu
Sven Kernke Voice input to text to fill in your input prompt box instead of typing it out! you will have soon, we are working on it .. Read out loud also we can do but we not doing rn ..

endu
updated the status to
planned

endu
Hey! Totally hear you on this—live speech is a bit tricky right now since it depends on OAI expanding their limits, but we’ve got TTS on the way! It’s actually part of our roadmap, so stay tuned—it’s coming soon! 😊
R
Radosław Domański
endu hey. It's actually super simple to implement using a local (browser) engine on the web and the system engine on the native app
If you'd like to - I can help you out on this, I have example code for both of those cases. It's literally a few hours of work to make it happen
that being said - for myself I just use this: https://dictanote.co/voicein/
works flawlessly

endu
Radosław Domański: Hey, thanks for sharing! 😊 Yeah, we've tried similar approaches, and while it's definitely doable, we found the voices to still sound a bit robotic, don’t you think? Would love your thoughts if you've found a way around that!"

endu
Alright, I’ll check them out and let you know how it goes. Thanks for the suggestions!
R
Radosław Domański
endu I was talking about the speech to tech part. as far as I'm concerned - that's a separate topic than TTS. And a much more powerful one. TTS is nice, but not required. So my previous answer was geared towards that.
but also - it's better to have a slightly robotic voice for free than none ;)

endu
Radosław Domański: Ahh, got it !