Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Nontransparent example at 231kbps (Read 11998 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Re: Nontransparent example at 231kbps

Reply #25
Also, your signature:
Quote
opus 96-192Kbps and aac-lc 96-128Kbps
Source 32bit floating point is fine don't need dithering
Source Fixed Point aka Integer should use dithering to prevent truncation distortion,
Why? 1bit Sign x 8bits exponent x 23Bits mantissa

My source file is indeed 32-bit float, generated by Adobe Audition.
X

Re: Nontransparent example at 231kbps

Reply #26
Killer samples are accepted. In this case it's rather a synthetic/generated sound — and these kind of samples are more a proof of concept than a real annoyance for users.

This type of thinking needs to be stopped. Opus (and most other lossy codecs) are designed for audible signals, not certain types of music. If it’s audible to humans, the codec should handle it. If that’s not the case, can i see the list of “acceptable noises to use in music”?

Re: Nontransparent example at 231kbps

Reply #27
On one hand yes and on the other no. Sure it would be nice to have a VBR that boosts the bitrate as much as necessary & sufficient  (but not much more) for no human ear to hear the difference - but if that takes so much that you might as well go lossless, then ... go lossless.
As long as you accept that these lossy formats are implementing a model that tries to trade off between size and quality, then there will be a wide range of bitrates where you can take a thousand signals, sort them into transparent and intransparent, rank the latter according to how obvious the artifacts are - and surely you can in the end, scroll to the bottom of that ranking list.

Re: Nontransparent example at 231kbps

Reply #28
Also, your signature:
So dither or not is irrelevant, if not worse.
if not worse probably yes
But your files are CBR right?
It's CVBR (Constrained VBR) and CBR

My source file is indeed 32-bit float, generated by Adobe Audition.
The WMA files were encoded using Adobe Audition 1.5's built-in encoder, with VBR settings.
it looks you are misunderstood, I wrote source it means RAW files (pure by hardware not converted, not modified or compressed by any built-in recording software), base on your comment I guess your original source is not recording in 32-bit float because I doubt the tone file just 2MB seems impossible.
32-bit floating-point use a high resource and wasted storage usage. Required capable hardware, possibly the tone file likely could take up hundreds of Megabytes. Consumer users shouldn't mess with that, easy to reduce SSD TBW lifespan unless they know what are they doing
maybe I should add detail to my signature

Killer samples are accepted. In this case it's rather a synthetic/generated sound — and these kinds of samples are more a proof of concept than a real annoyance for users.
This type of thinking needs to be stopped. Opus (and most other lossy codecs) are designed for audible signals, not certain types of music. If it’s audible to humans, the codec should handle it. If that’s not the case, can i see the list of “acceptable noises to use in music”?
I totally agree
Opus 96-192Kbps and Aac-lc 96-128Kbps
Source 32bit floating point is fine don't need dithering (dB noise -758dBFS that's a lot)
Source Fixed Point aka Integer should use dithering to prevent truncation distortion
Source is RAW

Re: Nontransparent example at 231kbps

Reply #29
Also, your signature:
Quote
opus 96-192Kbps and aac-lc 96-128Kbps
Source 32bit floating point is fine don't need dithering
Source Fixed Point aka Integer should use dithering to prevent truncation distortion,
Why? 1bit Sign x 8bits exponent x 23Bits mantissa

My source file is indeed 32-bit float, generated by Adobe Audition.
[attach type=image]21231[/attach]

hmmm... you might need to read an article about it ;)
https://www.sounddevices.com/sample-32-bit-float-and-24-bit-fixed-wav-files/
Opus 96-192Kbps and Aac-lc 96-128Kbps
Source 32bit floating point is fine don't need dithering (dB noise -758dBFS that's a lot)
Source Fixed Point aka Integer should use dithering to prevent truncation distortion
Source is RAW

Re: Nontransparent example at 231kbps

Reply #30
hmmm... you might need to read an article about it ;)
https://www.sounddevices.com/sample-32-bit-float-and-24-bit-fixed-wav-files/
Before you believe you have a "very" deep understanding of floating point by merely pointing to that Sound Devices article, better read this first:
https://www.audiosciencereview.com/forum/index.php?threads/zoom-f6-portable-field-recorder-review.15668/post-507610

The fact is floating point is merely marketing nonsense in the context of these field recorders, and the real world measured performance is pretty poor.
https://www.audiosciencereview.com/forum/index.php?threads/zoom-f6-portable-field-recorder-review.15668/

Even after updating firmware:
https://www.audiosciencereview.com/forum/index.php?threads/zoom-f6-portable-field-recorder-review.15668/post-786854

Also, try my software:
https://hydrogenaud.io/index.php?topic=114816.msg992983#msg992983

To conclude, you simply used a different encoding scheme (CVBR) to reduce the bitrate, and the whole thing is irrelevant to dithering at all, in the context of my posted test files.

 

Re: Nontransparent example at 231kbps

Reply #31
hmmm... you might need to read an article about it ;)
https://www.sounddevices.com/sample-32-bit-float-and-24-bit-fixed-wav-files/
Before you believe you have a "very" deep understanding of floating point by merely pointing to that Sound Devices article, better read this first:
https://www.audiosciencereview.com/forum/index.php?threads/zoom-f6-portable-field-recorder-review.15668/post-507610

The fact is floating point is merely marketing nonsense in the context of these field recorders, and the real world measured performance is pretty poor.
https://www.audiosciencereview.com/forum/index.php?threads/zoom-f6-portable-field-recorder-review.15668/

Even after updating firmware:
https://www.audiosciencereview.com/forum/index.php?threads/zoom-f6-portable-field-recorder-review.15668/post-786854

Also, try my software:
https://hydrogenaud.io/index.php?topic=114816.msg992983#msg992983

To conclude, you simply used a different encoding scheme (CVBR) to reduce the bitrate, and the whole thing is irrelevant to dithering at all, in the context of my posted test files.
thanks you anyway
Opus 96-192Kbps and Aac-lc 96-128Kbps
Source 32bit floating point is fine don't need dithering (dB noise -758dBFS that's a lot)
Source Fixed Point aka Integer should use dithering to prevent truncation distortion
Source is RAW