kalomaze
kalomaze
When converting a long file using Harvest on a Google Colab, it will visually 'error out' on the gradio interface, but in the TEMP path, the conversions eventually finish and...
There could be an advanced setting that detects where silence happens based on a threshold, and does each inference of long vocal pieces separately, then properly combines it into one...
Considering 40khz sounds better than 48khz, is this because upsampling introduces duplicate data in the file that, algorithimically, makes the output voice result sound 'more robotic' despite being trained at...
Config.json, when generated, should dynamically detect how long the dataset is, and set a faster logging rate in the config as necessary. Either that, or just use a much faster...
I notice 16khz seems to be hard programmed into the code for the downsampling phase. I think supporting more modular sample rates for inference audio would make sense. With librosa...
Are you guys using librosa.griffinlim or some other technique when converting to audio from spectrogram data?
I have been directing users in my community to only bother saving the .index file beginning with added_ Is it relevant to also have the trained_ .index file which is...
Setting 44100 as an option for the resampler causes only the first conversion to work. After that, it speeds up the audio to match sample rate requirements. Doesn't work as...
I was wondering if the batch processing feature on the interface processes all files at once by doing threading (since pm and harvest are on the primarily CPU), or each...
Sometimes HUBERT mishears words (phonetics?) and transcribes them incorrectly. Is there a potential solution where you can manually write what is being fed when vocoding?