Talk to ChatGPT: Voice to Voice

preview_player
Показать описание
Talk to ChatGPT: Voice to Voice

🗣️ Get Talk to ChatGPT here:

👊 Become a member:

Get a FREE 45+ ChatGPT Prompts PDF here:
📧 Join the newsletter:

🌐 My website:

In this video i create a simple Python script that lets you talk to ChatGPT, voice to voice. Pretty cool, and you can download the script on GitHub.

00:00 ChatGPT - Voice to Voice Intro
00:20 How ChatGPT Voice to Voice Works
01:56 ChatGPT Voice to Voice Demo
05:20 Get Your API Keys
Рекомендации по теме
Комментарии
Автор

Unless you are streaming (which is not the case) you can't get that fast response as shown in the video. There is a delay from saving the recording, transcribing it, sending to openAi, getting the response, converting to speach (tts), saving the audio file and playing it. This video must have been edited to remove the delays and make it look like a normal conversation. The content and the code are good and is definitely worth it (I wish I had that about a month ago when I built my talking gpt, would have saved me time), but there is no need to make it look like it's better than it is.

joaocarlosleme
Автор

I can tell you edited the video to show us immediate responses from chatgpt. It still looks great and shows what's coming.

jafetmorales
Автор

Curious to know if you like/recommend this over the Chrome Extensions "Talk-to-ChatGPT", "Voice Control for ChatGPT", or "VoiceWave: ChatGPT Voice Control"? I personally prefer the last one.

ronbates
Автор

Hello everyone and @allaboutai thanks for the code. Im very new to Github. Did I missed the part where I take the cod from git up and add it to gpt4? or do we just need the API key? probably a remedial questions. maybe I need to watch a git up tuturial first lol

pasqualenocito
Автор

This is awesome. I was just trying to get this exact thing working.

NightSpyderTech
Автор

I love this and use it a lot, my only request would be to have some sort of push to talk, or pause function. I'd love to have an ongoing conversation, but sometimes I need to focus on something else, and when I don't respond it usually just starts responding to the silence. (in Japanese for some reason!?)

captainbryan
Автор

Hey, this was an awesome video, but how much trim have you made, like how much time will it response before you edited the video

MidnightMysteries-pzhr
Автор

Nice! In my case I had to set channels=1. It also worked will in German. I've switched to gpt-4-0613 and used `eleven_multilingual_v1` as model_id. Also translated prompt to German, of course.

kostik
Автор

Is there a word limit for 11 Labs? Do I need to pay for an upgrade after I reach a certain number of words?

jojoen
Автор

Works great! I would love to see a version that works through a GUI. That has been really hard to transition to as well since I get messages/errors about the difficulty of working with the microphone and/or audio files outside of the python terminal.

NightSpyderTech
Автор

It would be great to have a detailed version. I have several questions. I have the API keys but I dont know which files and where to put them and where to call this up. Most here probably get that but I was left behind. I love your videos though!

athyrugbyclub
Автор

I got the voice bot working. Can I now add task completion functionality?

aitruthbyace
Автор

Awesome - I am the 1000th person to like it !!! Shows how outdated I am, but there is a start. Thank you for bringing this to us.

amitavamitra
Автор

Perfect, now... I had some problems with Voice ID and permissions to create temporary files, but now it's okay. ChatGPT saved me. Thanks, Kris!

FelipeVilas-rldb
Автор

Failed to run it on MacOS Ventura 13.5. Looks like sounddevice compatible problem for the MacBook Pro?

willmori
Автор

Is there a way to have it use a different voice every time? Like a random voice or something like that?

Kimari
Автор

what kind of prices are we looking at? it seems pretty cool.

homeofficeenglish
Автор

I've got a Question regarding using Elevanlabs API key. With the speech synthesis on the original page you have credits (quota) that vary depending on the subscription. Is the use of the API key for TTS in the program then also debited to the account, or is the number of spoken parts unlimited?

andrem.
Автор

I am constantly getting PermissionError: [Errno 13] Permission denied when doing the text_to_speech. I changed the temp folder, used Python as an administrator, asked ChatGPT for alternatives, and definitely cannot get it to play the response audio 😢

geronimotutusaus
Автор

Hope to see a tutorial that user speeches are turned into texts in chat box before AI answer. Thereby, I can see if I had any misstatement or dictation mistakes. Particularly useful for foreigners.

andyle