There's a thread title "How many fake file you have on your computer?" with reference to the application Fakin' The Funk?

I've just stumbled on another application:
Similarity - looks technically superior!
https://www.similarityapp.com/

Does anyone use this or have experience with it?

 

This seems like abandonware: last version is from 2019. Anyway, just tried: it works

 

I use Spek

 

Also - if there are 2 versions of a same mp3 track, both a same bit-rate but one in CBR one in VBR.
- which one to choose to keep?

 

If were talking MP3 here then variable bit rate files are generally more degraded than the constant bit rate files (ONLY if the CBR file is of decent KBPS). However if the CBR file is lower than 128KBPS then it might prove better to keep the VBR file. I don't play these games anymore with MP3 because it's just too dam artifact-y for my tastes. I dropped MP3 over a decade ago and started using FLAC exclusively and have never looked back.

 

Yeah - and both MP3 & FLAC can be encoded with various quality settings. So for the OCD of anyone, things can be nitty-gritty :-/

 

In a Flac file, compression affects size of file and is not at all related to quality. In a certain way, you can consider a Flac file like a zip format for audio

 

Or simply Lossless vs Lossy audio. If you leave a streamripper as "original format" it will rip straight to .flac. Changing the file format makes it take longer. Faster to damage your files locally, if you really want to go that route.

If you care about what a "Fake files" program says, you might want to consider just paying for them in the first place.

 

I would like to open an off topic, (but not so much) about YouTube audio compression.

On YouTube there are video files with an audio track that reaches up to 20khz (lossy), most of them instead, have the usual rolloff at 16k. I studied the issue and it seems that it mainly depends on the views: the platform saves the original uploaded file, when the algorithm notices that the views increase, the audio is re-encoded and the compression is improved. This was an in-depth study I had done some time ago to force my videos to be encoded with good resolution: this it is not an option that can be controlled by the user, it is the number of views that determine this process.
I do not have any source to link on the topic, but if I find the article I will gladly do so, I think it was AVS forum. Unfortunately, youtube is not transparent on their encoding/streaming procedures (no idea why)

 

Interesting.
So you found that yt dynamically re-encodes the original vs views?
That must use a lot of resource power!

It defeats the purpose of a title claiming 'HQ' etc.

 

The HQ in the title doesn't mean anything: the user has no control over how the content is encoded. ̶I̶t̶ ̶s̶e̶e̶m̶s̶ ̶t̶h̶a̶t̶ ̶t̶h̶e̶ ̶a̶l̶g̶o̶r̶i̶t̶h̶m̶ ̶i̶s̶ ̶b̶a̶s̶e̶d̶ ̶o̶n̶ ̶t̶h̶e̶ ̶a̶c̶c̶e̶l̶e̶r̶a̶t̶i̶o̶n̶ ̶o̶f̶ ̶v̶i̶e̶w̶s̶:̶ ̶f̶o̶r̶ ̶e̶x̶a̶m̶p̶l̶e̶,̶ ̶i̶f̶ ̶y̶o̶u̶ ̶h̶a̶v̶e̶ ̶1̶0̶0̶k̶ ̶i̶n̶ ̶3̶ ̶d̶a̶y̶s̶,̶ ̶y̶o̶u̶ ̶w̶i̶l̶l̶ ̶h̶a̶v̶e̶ ̶r̶e̶-̶e̶n̶c̶o̶d̶i̶n̶g̶.̶ ̶I̶ ̶h̶a̶v̶e̶ ̶v̶i̶d̶e̶o̶s̶ ̶i̶n̶ ̶m̶y̶ ̶a̶c̶c̶o̶u̶n̶t̶ ̶w̶i̶t̶h̶ ̶1̶0̶0̶k̶ ̶b̶u̶t̶ ̶t̶h̶e̶y̶ ̶w̶e̶r̶e̶ ̶g̶e̶n̶e̶r̶a̶t̶e̶d̶ ̶i̶n̶ ̶1̶0̶ ̶y̶e̶a̶r̶s̶,̶ ̶s̶o̶ ̶n̶o̶ ̶r̶e̶-̶e̶n̶c̶o̶d̶i̶n̶g̶ ̶a̶n̶d̶ ̶t̶h̶e̶ ̶a̶u̶d̶i̶o̶ ̶i̶s̶ ̶t̶h̶e̶ ̶u̶s̶u̶a̶l̶ ̶1̶6̶k̶ ̶s̶h̶i̶t̶.̶ If anyone knows more about the subject I would like to talk about it.
[edited]

Note: subscribed users have better audio on youtube videos

 

Last edited: May 31, 2025 at 6:12 AM

That is correct and incorrect. The audio that rolls off at ~16khz is m4a audio that is bundled with mp4 via CDN. The audio that rolls off at 20khz is webm (opus audio), that is bundled with webm (vp9 video). It is not dependent on views, Youtube will always offer both options because it is part of the encoding process. You can check that by parsing "-f 140" and "-f 251" via yt-dlp in command line (pip install yt-dlp), those will return m4a and webm audio respectively. I double-checked with a video from 11 days ago with 2.5 million views, and another one I had uploaded the same day with only 8 views. The audio and video you get is dependent on the platform you're using, Youtube will serve the video differently based on the device you're using.

Edit: you also added that subscribers have better audio, which is also incorrect. I pulled both audios from a technically logged off account, since I was lazy to export cookies to check the Youtube Premium-only resolution (it is video-only, so it wouldn't change anything).

Edit 2: further clarified webm information

Edit 3: data below

ID EXT RESOLUTION FPS CH │ FILESIZE TBR PROTO │ VCODEC VBR ACODEC ABR ASR MORE INFO
249 webm audio only 2 │ 1.60MiB 45k https │ audio only opus 45k 48k [ja] low, webm_dash
250 webm audio only 2 │ 2.10MiB 60k https │ audio only opus 60k 48k [ja] low, webm_dash
140 m4a audio only 2 │ 4.57MiB 129k https │ audio only mp4a.40.2 129k 44k [ja] medium, m4a_dash
251 webm audio only 2 │ 4.12MiB 117k https │ audio only opus 117k 48k [ja] medium, webm_dash
602 mp4 256x144 12 │ ~ 3.43MiB 97k m3u8 │ vp09.00.10.08 97k video only
269 mp4 256x144 24 │ ~ 6.07MiB 172k m3u8 │ avc1.4D400C 172k video only
160 mp4 256x144 24 │ 1.93MiB 55k https │ avc1.4d400c 55k video only 144p, mp4_dash
603 mp4 256x144 24 │ ~ 5.53MiB 157k m3u8 │ vp09.00.11.08 157k video only
278 webm 256x144 24 │ 2.22MiB 63k https │ vp9 63k video only 144p, webm_dash
394 mp4 256x144 24 │ 1.81MiB 51k https │ av01.0.00M.08 51k video only 144p, mp4_dash
229 mp4 426x240 24 │ ~ 12.38MiB 351k m3u8 │ avc1.4D4015 351k video only
133 mp4 426x240 24 │ 3.48MiB 99k https │ avc1.4d4015 99k video only 240p, mp4_dash
604 mp4 426x240 24 │ ~ 10.49MiB 297k m3u8 │ vp09.00.20.08 297k video only
242 webm 426x240 24 │ 3.52MiB 100k https │ vp9 100k video only 240p, webm_dash
395 mp4 426x240 24 │ 3.23MiB 92k https │ av01.0.00M.08 92k video only 240p, mp4_dash
230 mp4 640x360 24 │ ~ 26.26MiB 744k m3u8 │ avc1.4D401E 744k video only
134 mp4 640x360 24 │ 6.87MiB 195k https │ avc1.4d401e 195k video only 360p, mp4_dash
18 mp4 640x360 24 2 │ 17.18MiB 487k https │ avc1.42001E mp4a.40.2 44k [ja] 360p
605 mp4 640x360 24 │ ~ 22.31MiB 632k m3u8 │ vp09.00.21.08 632k video only
243 webm 640x360 24 │ 8.07MiB 229k https │ vp9 229k video only 360p, webm_dash
396 mp4 640x360 24 │ 6.20MiB 176k https │ av01.0.01M.08 176k video only 360p, mp4_dash
231 mp4 854x480 24 │ ~ 44.33MiB 1256k m3u8 │ avc1.4D401E 1256k video only
135 mp4 854x480 24 │ 10.88MiB 308k https │ avc1.4d401e 308k video only 480p, mp4_dash
606 mp4 854x480 24 │ ~ 41.39MiB 1173k m3u8 │ vp09.00.30.08 1173k video only
244 webm 854x480 24 │ 11.65MiB 330k https │ vp9 330k video only 480p, webm_dash
397 mp4 854x480 24 │ 10.06MiB 285k https │ av01.0.04M.08 285k video only 480p, mp4_dash
232 mp4 1280x720 24 │ ~ 82.08MiB 2326k m3u8 │ avc1.4D401F 2326k video only
136 mp4 1280x720 24 │ 20.23MiB 573k https │ avc1.4d401f 573k video only 720p, mp4_dash
609 mp4 1280x720 24 │ ~ 74.21MiB 2103k m3u8 │ vp09.00.31.08 2103k video only
247 webm 1280x720 24 │ 21.41MiB 607k https │ vp9 607k video only 720p, webm_dash
398 mp4 1280x720 24 │ 17.60MiB 499k https │ av01.0.05M.08 499k video only 720p, mp4_dash
270 mp4 1920x1080 24 │ ~168.88MiB 4786k m3u8 │ avc1.640028 4786k video only
137 mp4 1920x1080 24 │ 60.97MiB 1728k https │ avc1.640028 1728k video only 1080p, mp4_dash
614 mp4 1920x1080 24 │ ~121.46MiB 3442k m3u8 │ vp09.00.40.08 3442k video only
248 webm 1920x1080 24 │ 35.57MiB 1008k https │ vp9 1008k video only 1080p, webm_dash
399 mp4 1920x1080 24 │ 32.69MiB 927k https │ av01.0.08M.08 927k video only 1080p, mp4_dash
616 mp4 1920x1080 24 │ ~204.59MiB 5798k m3u8 │ vp09.00.40.08 5798k video only Premium

 

Last edited: May 31, 2025 at 5:46 AM

it's been a while since I downloaded audio from YouTube: I tested now and yes, they're all opus 20khz: great!! They must have improved the algorithm.

Regarding youtube premium audio, read here:
“Premium members will experience new audio enhancements, enjoying high-quality 256kbps audio on music videos with more clarity and depth”
Source: https://blog.youtube/news-and-events/new-youtube-premium-features-2025/

 

I use the free CCleaner " Remove Duplicates " !

 

Isn't Youtube a sort of "platform of last resort" for tracks you just cannot find anywhere else?

 

I use massively Jdownloader, and from a point up opus 152k audio disappeared, you see m4a as the first audio stream.
I noticed that until few months ago, in most yt clips opus 152k as audio stream was in most cases available, but it's not anymore.
When you listen to the video the cut is about 20kHz, so it's embedded in the video, but you have to extract from it.
However some video still has the opus 152k, this is today test:

 

Yes, but will do.

 

Some time ago I did a small test using the same flac reference, here the results:
Original:

File Converter 2.0.2 (context menu converter - FFMPEG based)
192k CBR:

Other commercial converter
192k CBR:

Same at 190k VBR:

Same at 192k ABR:


VBR and ABR were marked as fake.

 

I downloaded some tutorial on sister site a while back about EDM vocal production, and this guy's first chapter video was about how to rip stuff from YT to steal vocals from songs, or how to find them with the vocals isolated. His solution to the obvious problems was "having a real vocalist re-record it" if the quality is too degraded, and hinted at copyright evasion being another reason.

I thought that was terrible advice and deleted his "tutorial" without watching it any further. The one thing he did have right though, was how often something obscure enough that you would want to sample it, means it often is only on Youtube, or if older music maybe as an mp3 on SLSK. There is literally 1 song I would like to make a personal edit of, which is that exact situation. Or by ordering a CD copy of it from Europe, as it was an import-only in the US. My more generic question is then, when do you draw the line and give up and YT rip something like that?

Or are people using YT ripping just for stuff to listen to on phone mp3 player, car, etc?