search

SociallyIneptWeeb / LanguageLeapAI

Your Personal Multilingual AI Translator
MIT License
844 stars 169 forks source link

Voice sounds incredibly robotic #44

Closed QuiteAFancyEmerald closed 1 year ago

QuiteAFancyEmerald commented 1 year ago

Hello,

I was just curious on if you had a potential fix for preventing this from sounding incredibly robotic. I had resorted to tweaking the settings below but for whatever reason it clips at points (despite having a good mic although I don't think this is the reason) and just sounds off compared to your initial video.

VOICE_ID=20

# Type: float
SPEED_SCALE=1.0
VOLUME_SCALE=3.0
INTONATION_SCALE=1.5
PRE_PHONEME_LENGTH=1.0
POST_PHONEME_LENGTH=1.0
SociallyIneptWeeb commented 1 year ago

I'm not sure what you mean by clips, but if it sounds robotic maybe you can try different voices, also the quality of your mic doesn't matter as the TTS voice is simply being played into your speakers and into the vac cable

QuiteAFancyEmerald commented 1 year ago

I could send an example of what I mean by "clipping" but I'll try that. I just am curious how the same voice sounds so much better in your video or if maybe the settings above are not optimal?

https://cdn.discordapp.com/attachments/667692213022752769/1084311838281441280/ai_moments.mp4

SociallyIneptWeeb commented 1 year ago

From the sounds of it, it just sounds a bit fast, so you just have to use a speed scale of like 0.8 or something. Also idk why there are error messages appearing even tho the voice is properly being played

QuiteAFancyEmerald commented 1 year ago

Yeah its all good. Lowering the speed seemed to help. As for the errors no clue. I am using Google Collab with the proper environmental settings and have python 3.10.10 installed. Perhaps this is just a Google Collab thing since I have yet to try out Docker locally.

Amazing project btw