tacotron2 VS tortoise-tts

Which might use https://github.com/NVIDIA/tacotron2 in their backend

One thing is uberduck.ai but I think it's freemium (it's free but some features are premium). There's also tacotron 2.0 and its pytorch page. Many other softwares on sub but tacotron gave this and this and this.

Maybe we can use AI to hear this rapped by a famous rapper?

Sadly GothicBot the TTS I knew, doesn't exist anymore, but here is an alternative. It works in polish from what I heard.

From what I’ve seen, NVIDIA’s Tacotron2 can already be used to create some pretty convincing singing.

You might want to try something like Tacotron 2 by Nvidia to experiment with your current data.

We have the TorToiSe repo, the SV2TTS repo, and from here you have the other models like Tacotron 2, FastSpeech 2, and such. A there is a lot that goes into training a baseline for these models on the LJSpeech and LibriTTS datasets. Fine tuning is left up to the user.

I had tried tacotron2 + waveglow and it's quite easy to get very good results. The hardest part is collecting clean data.

I use Tortoise TTS. It's slow, a little clunky, and sometimes the output gets downright weird. But it's the best quality-oriented TTS I've found that I can run locally.

https://github.com/neonbjb/tortoise-tts

neonbjb/tortoise-tts

EDIT: It appears to be the exact same issue as this

It was primarily being used to train TTS models (see https://github.com/neonbjb/tortoise-tts), which largely fit into a single GPUs memory. So, for data parallelism, x8 PCIe isn't that much of a concern.

Previously TortoiseTTS was associated with PlayHT in some way, although the exact connection is a bit vague [0].

From the descriptions here it sounds a lot like AudioLM / SPEAR TTS / some of Meta's recent multilingual TTS approaches, although those models are not open source, sounds like PlayHT's approach is in a similar spirit. The discussion of "mel tokens" is closer to what I would call the classic TTS pipeline in many ways... PlayHT has generally been kind of closed about what they used, would be interesting to know more.

I assume the key factor here is high quality, emotive audio with good data cleaning processes. Probably not even a lot of data, at least in the scale of "a lot" in speech, e.g. ASR (millions of hours) or TTS (hundreds to thousands). As opposed to some radically new architectural piece never before seen in the literature, there are lots of really nice tools for emotive and expressive TTS buried in recent years of publications.

Tacotron 2 is perfectly capable of this type of stuff as well, as shown by Dessa [1] a few years ago (this writeup is a nice intro to TTS concepts). With the limit largely being, at some point you haven't heard certain phonetic sounds before in a voice, and need to do something to get plausible outcomes for new voices.

[0] Discussion here https://github.com/neonbjb/tortoise-tts/issues/182#issuecomm...

[1] https://medium.com/dessa-news/realtalk-how-it-works-94c1afda...

Tortoise GitHub repo - Source code, documentation, and usage guide

TorToiSe (https://github.com/neonbjb/tortoise-tts) produces the best quality speech of any freely available model. However, its long inference times makes it impractical for voice chatbots like Gdansk.