3 ms·
Try https://github.com/neonbjb/tortoise-tts https://github.com/neonbjb/tortoise-tts
by mlboss 4y ago
Try https://github.com/neonbjb/tortoise-tts https://github.com/neonbjb/tortoise-tts
- UncleEntity 4y agoThat one requires a big GPU and isn’t real-time. If you want to clone a voice and have a shitton of compute to fine tune it’s a good one. If you just want your computer to tell you you need to be out the door in 30 seconds or you’ll miss the bus then not so much.
- dr_kretyn 4y agoI found that it's much easier for me to read and remember when reading with voice assistant for which I need real-time. Ages ago I bought Ivona Text-to-speech and was serving me very well for many years. The last few years I used AWS Polly and espeak (using this https://github.com/laszukdawid/cracker https://github.com/laszukdawid/cracker) but thought that there must be something better.
- UncleEntity 4y agoThere seems to be a fairly wide selection between state of the art and just glueing together a bunch of phonemes it’s just that tortoise-tts is up there with the state of the art. I haven’t looked into the mid range stuff but there’s probably something out there with pretty good quality if you don’t mind doing some coding, end user applications seem to be mostly in the startup SaaS charge by the character domain.
- dr_kretyn 4y agoThanks! That was the first search and has nicely written colab, so will definitely give it a try. However, I've seen in readme that generating a sentence takes quite a long time. > On a K80, expect to generate a medium sized sentence every 2 minutes. Are you aware of other available?