Instant Audio Streaming with ElevenLabs AI Voice API - Here's How
Вставка
- Опубліковано 28 кві 2023
- github.com/jsammarco/ElevenLa...
Check out ConsultingJoe.com for more.
~-~~-~~~-~~-~
Please watch: "4K RC Flight in Shorewood, IL with Bixler 1.1 Plane & GoPro Hero 10"
• 4K RC Flight in Shorew...
~-~~-~~~-~~-~ - Наука та технологія
Great video! Thanks for sharing the tip.
Thanks for watching!
Awesome vid and just wonder: is there any way to use voice_id from Multilingual v2? I can tell that if I use a voice ID I trained with Multilingual v2, the sounds strange... Thanks.
Not quite sure. Sry, haven't used the api lately myself.
Hi Joe, do you know how to get the Elevenlabs Voices into a Twitch stream for the Community? I See a lot of streamen doing it for his donations.
I dont know much on twitch. Sorry
great video, just letting you know your api key and voice id are pushed to github
Thanks for that! Already rotated it.
Has anyone tackled streaming text in to then stream audio out; looking to see I can generate real-time audio as the generative text comes in. I appreciate some of the complexities with chunking and buffering, but every example I've seen to date sends into pre-defined text in to then stream out (which is great), but I really wan the audio to start as soon as I get text back from a LLM without waiting for the whole text to be completed.
Curious if you or the community has thoughts.
I have played with Ollama and created my own ChatGPT-like web interface but to stream to elevenlabs you would need to chunk the streams up and then there would be gaps in the speech. You could just break it up by paragraphs. Thanks for watching ❤
Thanks, @ConsultingjoeOnline. That's where I'm leaning, but for the obvious concerns with buffering and chunking. At least I'm not entirely crazy to think this way nor identify some of the concerns/issues. Ideally someone will provide a chunk'd text+audio stream simultaneously only as this would provide the lowest possible latency (I believe).
I ran some tests of this code and it turns out it's just as fast as using the non-streaming API :( If you want to really improve speed use ?optimize_streaming_latency=1 at the end of the URL it shaves off about 2 seconds without any noticeable drop in quality. You can also take it up to =4 but I found that the difference is negligible for my use case. But the default is 0 so using even 1 will get you speed results.
Very good ! I tested the code and it works; the only strange thing is that the audio has an English accent, in whatever language the sentence is to be transformed from text to voice.
As far as you know, do you need to enter any other parameters? Thanks for your answer ..
Thanks for watching. No that's weird. Try another voice with default parameters
This video is quite cool! Do you know why I am getting this error 401 code every time I try running it on my end?
Maybe issue with api key?
how does this work with Next.js?
i can't get the client side working to stream the audio.
did you sort it out? I have my own code in Python streaming from elevenlabs which I then relayed to a websocket so browser can consume but the audio is super choppy for some reason
@@thedoctor5478 hey did you figure how to do it? or please if possible could you share your code
Have you asked chat gpt about it?
Any example for Javascript? it would be great help
Like with NodeJS?
@@ConsultingjoeOnline actually node js is server side right? But I am talking about client side with venila javascript
hello, i have a problem, i get error 400 what do i do about it?
Hope you figured it out. www.google.com/search?q=error+400+meaning
Maybe bad API key or missing params