Local „ChatGPT“ Chatbot talk with LLaMA/GPT4ALL + Coqui TTS 🤯 | Install Tutorial

[@QuentinCretenet]

Discovered your channel yesterday, its brilliant, thank you for the quality content!

[@ThorstenMueller]

Thanks a lot for your amazing feedback 🤩.

[@svgsprite4942]

Your videos inspire me to continue my open-source project (prototext app). I can't wait to release the new version of my app, where users will be able to freely integrate any APIs, including local ones, not just ChatGPT, Whisper, and Elevenlabs. Your videos have helped me learn how to set up local neural networks. Thank you, Thorsten!

[@ThorstenMueller]

Thanks for you amazing feedback 😊. Good luck for your next app release.

[@magenta6]

Great clear tutorial Thorsten! "What a time to be alive!!"

[@ThorstenMueller]

Thanks Paul, and "yes" 😊.

[@nephisd1029]

I originally discovered your channel because I wanted to do exactly this and I started with CoquiTTS. I got that working, but didn't do the language model part. Can't wait to try it!

[@ThorstenMueller]

Thank you very much for your feedback. I'm really happy you find my content useful 😃.

[@jma42]

This is what I felt really after discovering LLaMA models thanks for this, I might actually start doing a dream project of mine LOL

[@ThorstenMueller]

You're welcome and good luck with your "dream project" 😊.

[@TheExplosion]

Thnaks! That will help a lot!

[@ThorstenMueller]

مرحبا صالح ، شكرا على ردود الفعل اللطيفة. (Thanks for your nice feedback).

[@UserB_tm]

Great video. I'm a fan of Nomic AI. They have a brilliant team.

[@ThorstenMueller]

Thanks for your nice feedback 😊.

[@toykotokyoto]

Stoked to see how this all can be done locally!

[@ThorstenMueller]

Absolutely - no need for external cloud services!

[@nobodynobody5283]

Das ist bombastisch!

[@ThorstenMueller]

Vielen Dank für dieses absolut tolle Feedback 🤩. Ich freue mich, wenn das Video hilfreich ist.

[@Anon-iz6hm]

Thank you for the great video. Can this setup work on Raspberry Pi. Is there any tutorial / steps around this?

[@ThorstenMueller]

You're very welcome 😊. I didn't try it yet, but as Raspberry Pi isn't mentioned on their Github page i guess it's not their main focus.

[@andrelaurin540]

Great use of TTS, but where did you get the GPT4-J model? I have tried the Nomic GPT4all models but they will not load in GPT4-J. I am a beginner in python and found your channel very intriguing. Thanks for all your effort. It would be nice if you used a bigger font also. The terminal screen in your video is very blurry.

[@ThorstenMueller]

Thanks for your feedback on my content and your suggestion to increase font size in terminal. I'll keep this in mind when making my next videos 👍. Finding the right model could be difficult. Do you get an error message when trying to load the model?

[@guilloisvincent2286]

Nice ! I needed that for my npc :D (trying to kill server response time hope it will be quick)

[@ThorstenMueller]

"Quick" is depending on available compute power - as always 😆.

[@guilloisvincent2286]

@@ThorstenMueller my pc is quite good with 2080, Hope it will be < 2sec

[@redfield126]

Excellent content. You have a new fan. Is coqui TTS open source ? I see model are flagged Apache 2 but not sure 100%

[@ThorstenMueller]

Thanks for your nice feedback, this is really appreciated 😊. Yes, Coqui TTS is open source (apart from some parts as their Studio) and each model might come with their individual license.

[@redfield126]

@@ThorstenMueller thanks again for your content. I followed your steps and it works like a charm. As you told,, that is crazy what we can achieve now on a consumer hardware. I’ll stay tuned for your next episodes !

[@fire17102]

Maybe instead of printing outText, try to run foreach line in outText: (or outText.split(" ") ) Then generate and play audio Could be useful to start reading while the gpt is still processing :) good luck & thanks for the videos!

[@fire17102]

On further inspection I see that the audio can be triggered to generate at new_text_callback. This should be asynchronous, and pushed into the audio playback queue

[@ThorstenMueller]

Thanks, that really a good hint 👍.

[@redbaron3555]

Super Video und Idee danke!! Leider ist der TTS Model Download broken (SSL Fehler). Auch manueller download erfordert ein hardcoded des Pfads....Haben Sie eine Idee wie man das lösen könnte? (MacAirM2) Python 3.11.03

[@ThorstenMueller]

Vielen Dank für das nette Feedback 😊. Ist eventuell ein Proxy/Filter, etc. im Spiel, der den Download blockieren könnte? In der Datei stehen die Original Download URLs drin. Klappt ein einfacher Browser Download? Eventuell hat der Download aber auch wirklich ein Problem, ist ja auch mal möglich. github.com/coqui-ai/TTS/blob/dev/TTS/.models.json

[@mohsenyazdani6984]

How about making it as voice to voice assistant?!! 🤔

[@jamiecallaghan2609]

Just edit his code and add a text to speech response using Coqui again!

[@philb1466]

🤯 ... I have no words!

[@ThorstenMueller]

I really hope, that's meant as a compliment 😉. If so, thanks a lot 😃.

[@philb1466]

😄 .. I always enjoy the content on your channel and very much appreciate you taking the time to make and share it! I'm just blown away at what is being achieved here! I'm a big fan of offline and only tether a computer to my phone when I want/need to go online, so, seeing this being done without having to be online is amazing!

[@magi987]

Hi, i really like your videos :) is it possible to combine coqui-TTS and coqui-sTT for Homeassistant? And if yes can you maybe make a video for it? Also it would be nice if you show how to do it with a docker containers?

[@ThorstenMueller]

Thanks for your nice feedback, i really appreciate it 😊. Do you know my video about using Coqui TTS in Home Assistant? kzbin.info/www/bejne/rKW9k3qof9aikLc On the STT side i'm not sure how Coqui STT compares to Whisper for example.

[@magi987]

@@ThorstenMueller yeah your CoquiTTS for home assistant I already watched that's why I ask because HA have this assistant feature and all I'm missing is local STT and a good guide for all of these in docker containers

[@ThorstenMueller]

@@magi987 I've a video for local STT in Home Assistant (2023.5) on my TODO list. So maybe this will show how to set this up.

[@magi987]

@@ThorstenMueller that sounds great I'm definitely looking forward to this :) thank you for making these unique videos all about STT and TTS

[@Asky_]

can it be integrated directly in the windows app? There you have the luxury to select the language model

[@ThorstenMueller]

You mean adding a Coqui TTS voice in Windows native?

[@fire17102]

Can you please share the minimum requirements to run this? Does it use the CPU or GPU? How much RAM needed. What specs where you using for this demo ? Looks fast enough

[@ThorstenMueller]

I'm not sure on the minimum requirements as there's nothing noted on github.com/nomic-ai/gpt4all. But i've used a MacBook Air M1 16GB RAM.

[@NickMak-m2c]

It's amazing but the SD end of AI has me really perceiving up the expanding potentials, as they say, for the someric positivatiols. But since realizing I could probably run this (switching our pronoun now to coqui) on an 8GB vram nvidia, I'm skeeting vehemently.

[@capitalcleaning]

You learn how to zoom or focus on the area you are talking about. Its a pain to look at the whole screen and the tiny letters. you are doing a good job but also defeating the purpose yourself. Thank you.

[@ThorstenMueller]

Thanks for your feedback and suggestion 😊. Maybe i will test some "zoom on mouse" extension on my next video.