mfcc64
mfcc64
It makes gain entry linear. The 15 bands placement is non-logarithmic (to allow 0 Hz) but at high frequency it becomes half-octaves band. Assuming t = 2.8853900817779269*log(f/157.48+1) Then f =...
Using cubic_interpolate(f) is enough. Note that the default is gain_interpolate(f) which is less smooth than cubic_interpolate(f).
Can you give me sample command and file to reproduce the bug?
I have no idea how to fix this.
major version pinning is not suitable for production - cache is too long with up to 14 days of cache
`max-age=604800, s-maxage=43200`, is this really intended? (7 days client cache, 12 hours CDN cache) Or should it be `max-age=43200, s-maxage=604800`? (12 hours client cache, 7 days CDN cache)
Does Chrome have similar result? Can you view page source of YouTube page and search for 'ShadyDOM'? Thank's.
That's different problem. Can you run this ```js const cqt = document.getElementsByTagName("showcqt-element")[0]; console.log("sample rate:", cqt.audio_context.sampleRate); console.log("dimension:", cqt.clientWidth, cqt.clientHeight); ``` Thank's.
Click the icon at extension sub menu. Click the top left icon on the page. Currently, this extension is unavailable on m.youtube.com, so probably you should change to desktop mode....
Can you give the dsf sample file? Thank's.
Feel free to make pull request. Thank's.