How to use the ElevenLabs API: Python text-to-speech tutorial with examples 

I really appreciate that, thank you!

That's great to hear - thanks!

Thanks Garrett! Like you, I was surprised there weren't more examples like this.

Cool, I'm delighted to hear that - thank you!

Pliss help sir

@@edosetiawan9589 not a great solution but I installed an older version in terminal (elevenlabs==0.2.27) and then I had to 'brew install ffmpeg' and then it worked!!

@@edosetiawan9589 i gave an complex answer. If you know something easier please post here

Hi Yaga. No problem, I think we can fix this. It sounds like you might have typed "pip install elevenlabs" directly into a text file (a Python program) and then tried to run it. Is that correct? If so, the "pip install elevenlabs" (without quotes) is something you type directly into the terminal *before* running your Python program. That will install the module. Then, in your program at the top, you'd type "import elevenlabs" to use the module. Does that make sense? If that doesn't work, it could be because your terminal is a kind of Python terminal (used for typing Python commands directly). In that case, type "exit()" (without quotes) to exit into a regular terminal. Then you can try "pip install elevenlabs" again. I hope that works! Let me know how it goes.

hi mate, I know that your message is 2 months old but I hope you will find this. Try instead typing this: python.exe -m pip install eleven labs

Nvm works the same way but require api key

@@railgundotenc8538 Hi there. Just in case, I made a list of all premade voices and their voice IDs in the description of this video: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-pXArMwOgjPU.html Hope that helps.

I'm talking about not premade voices.

@@railgundotenc8538 Ah sorry, my misunderstanding. Yes, as you mentioned, I think you need to use the API's "voices" endpoint for that: docs.elevenlabs.io/api-reference/voices

Hi again. In case it helps you (or others with the same question), I just made a video showing how to get custom voice IDs: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-r5aJeq-f0OY.html

I'm also curious to know if you can use an API for an instant cloned voice.

guys is this linux where can i code with python?!

Hi. Good question. There are a couple of ways (from the ElevenLabs docs or using Python) so I made a video showing how to do it: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-r5aJeq-f0OY.html I hope that helps!

Hi. That's strange. I've not come across this myself, but I found a few other people that had this problem and other people have suggested solutions. Perhaps one of these will work for you: github.com/elevenlabs/elevenlabs-python/issues/56 github.com/elevenlabs/elevenlabs-python/issues/59 github.com/elevenlabs/elevenlabs-python/issues/44 stackoverflow.com/questions/77150521/eleven-labs-api-not-working-is-ffmpeg-not-installed-properly If not, it'd be helpful to have some more details to help debug, e.g. what operating system you're using, whether you're using the terminal or a Python environment (e.g. Replit, Pycharm), how you installed ffmpeg (e.g. pip, brew), and what the full error message is.

I get the same thing :(

Hi. I just tried in PyCharm and I see what you mean. It seems that the VoiceSettings class doesn't work in PyCharm unfortunately (I don't know why), so here are a couple of workarounds: Option 1: Remove this from the example code: settings = elevenlabs.VoiceSettings( stability = 0, similarity_boost = 0.75 ) Option 2: Just use the voice name or voice ID directly in the generate() function, like this: audio = elevenlabs.generate( text = "Hi, I'm from the future!", voice = "Bella" ) or audio = elevenlabs.generate( text = "Hi, I'm from the future!", voice = "ZQe5CZNOzWyzPSCn5a3c" ) Either of these makes it work for me in PyCharm.

Hi Bill. Hmm, that's strange. For me it worked using "pip install elevenlabs" in the terminal (although I haven't tried on a Raspberry Pi). Maybe check that other packages can be installed as well, if you haven't already. For example, try "pip install openai" or something similar. One possible cause could be that both Python 2 and 3 are installed, in which case maybe "pip3 install elevenlabs" works instead? I hope that helps.

@@PuppyCoding I've done the pip and pip3 installs. Both work. It just errors if I try to specify the voice_id instead of the name.

@@BillYovino That's really weird - I don't understand why there'd be a difference. If you're able to paste that section of your code here (or somewhere else), I'm happy to take a look.

@@PuppyCoding import elevenlabs elevenlabs.set_api_key() voice = elevenlabs.Voice( voice_id = "vuZhMxquIvNVUPRorn2I", settings = elevenlabs.VoiceSettings( stability = 0.3, similarity_boost = 0.75 ) ) audio = elevenlabs.generate( text = "Hi, I'm from the future!", voice = voice ) elevenlabs.play(audio)

Hi again Bill. I did some investigating and it seems that the VoiceSettings class doesn't work in some software, e.g. PyCharm. I don't know why but maybe the the Raspberry 4 is the same. Anyway, here are a couple of workarounds: Option 1: Remove this from the example code: settings = elevenlabs.VoiceSettings( stability = 0, similarity_boost = 0.75 ) Option 2: Just use the voice name or voice ID directly in the generate() function, like this: audio = elevenlabs.generate( text = "Hi, I'm from the future!", voice = "Bella" ) or audio = elevenlabs.generate( text = "Hi, I'm from the future!", voice = "ZQe5CZNOzWyzPSCn5a3c" ) Either of these makes it work for me in PyCharm, so hopefully Raspberry Pi 4 as well.

Hi there - thank you! Yes, it is possible to use cloned voices with the ElevenLabs API, and thankfully it's straightforward - it's actually the same as with premade voices. You can access both premade and cloned voices with your API key. You just need to know the voice name or ID of the cloned voice. You can get the voice ID from the "voices" function of the API, either using Python code like this example here: github.com/elevenlabs/elevenlabs-python#%EF%B8%8F-voices Or using the Try It Out feature of the API docs here: api.elevenlabs.io/docs#/voices/Get_voices_v1_voices_get

@@PuppyCoding This helped a lot! Thank you so much for the extensive reply, you're a legend ahahaa

@@hedgedoge5562 Great - you're welcome!

Hi. I couldn't find an official page with the character limits, but in my testing I was able to use up to around 300 characters for free without an API key. For longer text it started asking me for an API key. If you need more than around 300 characters each time, then you'll need a paid account. Their pricing page is here: elevenlabs.io/pricing

Olá, Paolo. I don't think there's a way to choose a voice specifically for Portuguese unfortunately. I had the same problem with Japanese and the only thing I could do was go through each of the current ElevenLabs voices and see which one sounds best. I think we just have to wait for ElevenLabs to release voices for each language.

I found another alternative I used Microsoft Azure, it has voices in Portuguese,thanks for the feedback!

@@pauloroberto-gz4kv Oh nice, I didn't know that was possible. Thanks for letting me know.

Fala, Paulo! Tudo bem com você? Você achou um jeito de usar o TTS da Azure clonando a voz? Ou no caso deles só é possível utilizar com vozes já existentes? Outra dúvida: você não encontrou uma voz em português padrão com o ElevenLabs? Você considerou clonar alguma voz e usar ela? Muito obrigado! @@pauloroberto-gz4kv

That's true. I hope they add more languages soon because the quality of what they already have is so good.

Thanks for your comment. I agree, the way I code in the video can seem old fashioned, given the modern tools we now have available. I do it this way partly because I am old fashioned (sorry!) but also because my priority is for the videos to be educational for people just getting familiar with Python. I feel that typing everything by hand gives me the chance to explain what's going on, without code appearing "by magic". I should have made that more clear, so I've just updated the channel About page. Hopefully that reduces confusion, and thanks for pointing it out.

@@PuppyCoding you're fine, you can code however you want, i'm just surprised by how slowly AI tools are coming into use even on channels that talk a lot about AI ,, not just you, pretty universally, i guess that's just human speed, we're the very slowest thing that can general intelligence at all ,, have you tried copilot or asking gpt4 for programs yet yourself?? i think there's a huge opportunity coming up to allow people to ask for programs who aren't traditional style programmers,, & beginning programmers starting right now should be encouraged to leapfrog and just never learn the old way at all, that's my intuition

@@mungojelly Ah, I see what you mean. Yes, I think I agree, there doesn't seem to be that much hands-on content with AI programming tools. Perhaps the audience can be roughly divided - 1) those that want to learn how to build something, and 2) those that want to learn the fundamentals to embark on a career. Personally I prefer teaching the "old way" for people in the second group, but there is a need for content for the first group too. I do use ChatGPT + Code Interpreter (now called Advanced Data Analysis) for my own coding and love it, although I still prefer to verify its output before using it. Maybe I should do a video on that, from a more practical standpoint. Thanks for the food for thought!

Hi Kunal. You're right, you need a subscription to use the API long-term, but they allow a free initial trial without an API key, just to test it out. I found that I could use it up to about 300 characters for free. After that, it started asking me for an API key.

use Arnold instead of bella, bro if there is no bella just looks for other names on the webpage, you could have find out that urself

@@blueboi-wk8gq I tried all sorts of voices including Sarah, this library needs some complex 3rd party installation called FFmpeg, which of course is not explained by the uploader seeing as this "tutorial" is another classic example of search engine poisoning.