Any cheaper and better alternative to ElevenLabs?

[-]

Ritza-co@reddit

We've tried a few recently and wrote up a neutral comparison to Smallest here, focusing on real-time voice

https://techstackups.com/comparisons/smallest-ai-vs-elevenlabs/

[-]

Initial_Froyo4625@reddit

We switched to Gradium for most of the generation and it’s been a solid tradeoff on cost without the voice going full robotic, and voice cloning options are nice. I like fish as well.

[-]

Savings_Stress9988@reddit

Depends on what you’re building.

If you only need text-to-speech, there are a lot of alternatives. But if you're building full voice AI agents (like phone bots or AI receptionists), something like Retell AI might be a better fit.

It’s more of a voice-agent infrastructure than just a TTS provider.

[-]

opellec@reddit

I tested a few TTS recently, Fish audio feels about on par with Elevenlabs, just way cheaper.

[-]

herberz@reddit

try contextlm.ai

it has better quality voices at half the price

[-]

kangan987@reddit

I tried it and it sucks. It failed to generate the audio but still ate my credit.

[-]

ConstructionRough152@reddit

Anyone know anything better than Elevenlabs with RU please? I reached the max there also and more than 5000 characters I cannot do...

[-]

While there are some alternatives out there, ElevenLabs still stands out for quality and versatility. If budget is a concern, you might want to check out some voiceover tools that complement ElevenLabs’ free tier nicely—happy to share a link if you’re interested!

[-]

lucidnegro@reddit

Kindly share

[-]

Individual_Weird_685@reddit

Hello, I just created Amulet Voice and it's 80% cheaper than ElevenLabs, you can check it out!

[-]

Electrical-Cap7836@reddit

11Labs can get pricey fast. I used VoiceHub by DataQueue for voice in one of my projects (more focused on voice agents), but it worked well and was more cost-effective for my use case.

[-]

Electrical-Cap7836@reddit

11Labs can get pricey fast. I used VoiceHub by DataQueue for voice in one of my projects (more focused on voice agents), but it worked well and was more cost-effective for my use case.

[-]

IslamGamalig@reddit

I’ve been playing around with VoiceHub lately and it’s quite interesting to see how smoothly it handles text-to-speech workflows. Still experimenting to see how it compares to the bigger names, but so far it feels pretty solid.

[-]

Overall-Stuff7963@reddit

VocalCopyCat is significantly cheaper than ElevenLabs - their premium package is $100 for 10M characters vs ElevenLabs $1,650 for the same amount. Quality is comparable, sometimes better with fewer artifacts. We switched for our video product and cut voice synthesis costs by over 90%. You may have to reach out if you also want API access.

[-]

Overall-Stuff7963@reddit

VocalCopyCat is significantly cheaper than ElevenLabs - their premium package is $100 for 10M characters vs ElevenLabs' $1,650 for the same amount. Quality is comparable, sometimes better with fewer artifacts. We switched for our video product and cut voice synthesis costs by over 90%.

[-]

rbgo404@reddit

We have recently analysed a few open source TTS models. You can check them out here:
https://www.inferless.com/learn/comparing-different-text-to-speech---tts--models-for-different-use-cases

[-]

WompTune@reddit

I cannot click on yall's cookie banner. I've clicked both "Accept" or "Reject All" to no avail. Extremely annoying while trying to read what your website does

[-]

iamMess@reddit

I just added a free Kokoro TTS endpoint. It's not exactly ElevenLabs quality, but it comes really close.

Feel free to try it out: https://kokorotts.com - no strings attached. This community has given me so much, so just giving a little back.

[-]

Kindly-Annual-5504@reddit

It's not even close to elevenlab's quality. Elevenlabs plays in it's own league in terms of quality. XTTSv2 or some of its forks with some good quality speaker files could probably come very close. I used some files generated Wie elevenlabs as speecher files and it sounds really good. But it's not the fastest out there, but still decent.

[-]

Ok-Hedgehog-5086@reddit

It's not exactly ElevenLabs quality, but it comes really close.

Not even close. If it were a surface to air missile, it would hit empty air 100 km off target.

[-]

iamMess@reddit

You can also run it yourself. The model is quiet fast and small.

[-]

Widget2049@reddit

I uses ElevenLabs explicitly for JP voice, but I've replaced it with Tsukasa_Speech https://huggingface.co/Respair/Tsukasa_Speech. for examples you can use their interactive demo. i tried this awhile back https://gofile[.]io/d/UshCmC

[-]

Klutzy_Comfort_4443@reddit

Can you tell me if Tsukasa is better than Microsoft's TTS or ElevenLabs for Japanese? I've heard that ElevenLabs doesn't pronounce Japanese well. I think Microsoft's TTS is better, but it still has its issues.

[-]

Widget2049@reddit

strictly in japanese, IMO Tsukasa is better than Microsoft TTS because Tsukasa is also capable to add sigh, pauses, slight giggles, scoffs on some speaker.

If you're looking for something better, look TTS service that has SSML (Speech Synthesis Markup Language) included. that way you can truly finetune how the text will be read. I ever tried using IBM Watson TTS, the result is good. but well, it's IBM. in the end of the day it's still pricey and can't beat the locally hosted model. not to mention you also have to worry about your texts being send into someone's server if it contains "funny" stuff.

the thing about AI is that for time being they perform better under specific task. the more you can narrow that down, the better. Tsukasa has a leverage here because they're only focused in japanese language. so yeah, i think Tsukasa is better compared to Microsoft TTS and ElevenLabs

[-]