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: Replay gain Bug (Read 4105 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Replay gain Bug

Replay gain gives album gains of +48db to albums which peak at 0db.  This has happened to 24 & 32 bit wav's.  Also 32 bit floating point aiff's and some broadcast wav's do not play at all.

Replay gain Bug

Reply #1 –
Quote
Replay gain gives album gains of +48db to albums which peak at 0db.  This has happened to 24 & 32 bit wav's.

And what else did you expect on complete silence ?
Microsoft Windows: We can't script here, this is bat country.

Replay gain Bug

Reply #2 –
0 db's as far as I know equals the highest possible amplitude on non-floating point files, not silence.

Replay gain Bug

Reply #3 –
I'm sorry but your bug report doesn't make any sense to me - I have plenty of files with 1.0 (or "0dB") peak, all of them get correct values. Please upload files that trigger this behavior.
Microsoft Windows: We can't script here, this is bat country.

Replay gain Bug

Reply #5 –
Just downloaded your sample files and they replaygain scan just fine here.

The error message you get from the console simply means that you haven't got "Write APEv2 tags to WAV files" enabled in the input preferences.

In this case foobar will simply store the RG values in it's database instead of the file tags.  No real problem either and as long as you don't rebuild your database the files should remain replaygained.

Also neither of the files you uploaded have values anywhere near the astronomical +48 dB you mentioned.  It's more somewhere around  +10 dB.

All of this seems to be rather intended behaviour to me and I absolutely can't find any bugs here.

Replay gain Bug

Reply #6 –
I have uploaded 3 very short wav files as well as the log errors reported in 0.7.7a.  The problem has been constant over full length songs as well.

Thanks for your on-going efforts.

Replay gain Bug

Reply #7 –
Quote
I have uploaded 3 very short wav files as well as the log errors reported in 0.7.7a.  The problem has been constant over full length songs as well.

Thanks for your on-going efforts.

And what is your problem again ? These files most certainly don't get +48dB gain here.
Microsoft Windows: We can't script here, this is bat country.

Replay gain Bug

Reply #8 –
These are not the samples which get a replay gain of +48.  Instead they get an album gain of  +9.14db when it should be +0db as they are normalised already.  Also a error warning comes up when you analyse the files with replay gain.

Replay gain Bug

Reply #9 –
As I said

Quote
The error message you get from the console simply means that you haven't got "Write APEv2 tags to WAV files" enabled in the input preferences.

In this case foobar will simply store the RG values in it's database instead of the file tags. No real problem either and as long as you don't rebuild your database the files should remain replaygained.


It also does not surprise me that much that the files get a rather large replaygain value, since they consist of a very short burst of noise and a lot of silence.  Not exactly the best example for everyday use of replaygain.  Makes scence to me that the replaygain scanner gets a bit confused on a sample like this.  Someone correct me in case I am wrong though.

Replay gain Bug

Reply #10 –
- ReplayGain is NOT a peak normalizer. Please read this first.
- "Errors" you get occur because writing tags to WAV files is disabled by default, as picmixer pointed.
Please search the forum next time before posting false "bug reports", thank you.
Microsoft Windows: We can't script here, this is bat country.

Replay gain Bug

Reply #11 –
Quote
they get an album gain of +9.14db when it should be +0db as they are normalised already.

Should be "+0dB" according to what?
How did you normalize then, what method? what levels? with what application?

It sounds as if you are playing around with Audiotion or something and are a bit clueless about the options and setting therein.
PLEASE start to give some information with these "bug" reports - you're currently not making any sense and don't explain what is happening in an informative manner.
It's hard to theoretisize without proper explanations of the issues you present.

Quote
Also a error warning comes up when you analyse the files with replay gain.

Saying what?
How? In what application? Foobar?

Replay gain Bug

Reply #12 –
Normalising is a common DSP function where you set the maxium level of a file.  I chose to normalise to 0db, the highest level possible (in wavelab).  If you where to analyse files you would find that they are already at peak levels.  Where does to +9.14db come from?  All that is doing is making them clip by +9.14 (Bad).

Replay gain Bug

Reply #13 –
Quote
Normalising is a common DSP function where you set the maxium level of a file.  I chose to normalise to 0db, the highest level possible (in wavelab).  If you where to analyse files you would find that they are already at peak levels.  Where does to +9.14db come from?  All that is doing is making them clip by +9.14 (Bad).

The first thing you should do is start reading the replies handed to you above before posting out blindly like now.
Normalising can be done in several ways, with various methods..
You claim to have normalised to peak at 0dB - when you should do a maximum of .03 instead.

There must be - assuming you're correct - some sort of minipeak, such as a glitch from DAE failure or similar that is triggering WaveLab to correspond to this shortimpulse peak as maximum of 0dB whilst the rest lies at lower amplitude. ReplayGain bases on RMS value and peak value and could disregard this impulse when scanning. Hence the difference of +9.xdB

I still do not believe this is a bug of any sort, it still seems like a user error.
What options did you use when normalising in WaveLab?


In any case you should use Advanced Limiter for possible clipping salvation.