Ce forum permet à des personnes du monde entier de communiquer, c′est pourquoi les messages échangés sont en anglais.
-
Very slight noise is added to the sample every time I save the soundfont.
(version: polyphone 2.5.1)
Suppose you do "wav export" for one sample(for example, 400Hz sine wave), then apply some changes to the soundfont(anything but should not affect the sample itself. ex. add new preset) and save the soundfont. Close and open the soundfont and do "wav export" for the same sample you exported.
The two exported wav should be identical, but it's not. It shows difference if you compared by foobar bit compare tool(See the picture).
To see what type of noise was added, I imported them to Audacity, applied phase inversion to one sample, and mixed down them to one track. This will cancel the identical sound and show the difference between them. It seemed that pulse like very small noises were added.
(the last track shows the noise around - 90db)
This noise itself is not so problematic, because it is not audible. The problem is that it is accumulated every time I save the soundfont.
(The former: saved one time
The latter: saved several times
noise level was increased to -80 db)
I am afraid that this finally could affect the soundfont quality.
I have no idea why this kind of phenomena is happened. Is there very weak limiter or something??? -
Message de musiclikesoundfont leLet me add one more picture to show that the noise is actually pulse like.
And If the soundfont is in 24 bit, the noise is not pulse like but like white noise.
This is also problematic. This noise is too big for 24 bit, as the weak signal of 24 bit masked with the noise, and it will work as if 16 bit. -
Message de musiclikesoundfont leJust in case, original sine wave file used.
It's in flac file but 100% identical to the original wav file.400 24bit -1db.flac -
Message de musiclikesoundfont leIf the soundfont is not changed and saved, it exports exactly the same wav file, as it should.
-
why .flac when the issue is .wave?
-
Thank you for your reply.
To reduce the size of the file(because of 2MB file upload limitation).
I confirmed the flac file is identical with the original wav file (used in the experiment above) with foobar's bit compare tool.
So if you want to reproduce the phenomena, convert flac file to wav file and create new soundfont with the sample.
Then follow the same procedure I supposed above.
I hope it will work.
(If I have a time, I will try another way to upload the needed file to more directly show the phenomena.)
-
Message de musiclikesoundfont leOk, I uploed the outcome of the procedure.
The file contains 3 wav file. All of them were exported with polyphone's "wav export" function in the procedure described above.
The 3 files sounds the same to ear, but they are not exactly identical.
You can confirm it simply by using foobar2000's bit compare tool.
If there is any requirement and I have a time, I will add some explanation and file.
I think this phenomena is hard to confirm and explain, so I feel kind of nervous how I can explain.
I think there is a lot left to explain...16bit outcome of the procedure.zip -
i mixed the inverted first .wav and the third .wav in Soundforge and the result is an absolutely mute .wav file
-
bottrop, le -i mixed the inverted first .wav and the third .wav in Soundforge and the result is an absolutely mute .wav file
-
Message de musiclikesoundfont lebottrop, le -i mixed the inverted first .wav and the third .wav in Soundforge and the result is an absolutely mute .wav fileThank you for taking your time in confirming my report(Sorry to be late).
I am really glad if there is any solution for this(including my possible oversight or mistake), and if this post leads to some very small something to the developers. -
Message de musiclikesoundfont lebottrop, le -i mixed the inverted first .wav and the third .wav in Soundforge and the result is an absolutely mute .wav fileSorry to say that I do not own soundforge and I may not be able to reproduce your environment.
I used Audacity to invert, mix, and visualize. -
Message de musiclikesoundfont leEasiest way to check that the 3 files are actually not identical is to use foobar2000's "Bit compare tracks" function, which will show whether the sound signal itself is exactly the same(It ignore difference of file themselves(such as extension difference like .flac and wav). Purely compare the sound signal only).
Import the 2 file in foobar2000, select 2 files, and right-click. It will show the "Bit compare tracks" function as below (I think this function is built-in. There is no need to install plugin I guess. (foobar2000 v2.25.8 x64bit)).
Click "Bit compare tracks" then it will work.
-
Message de musiclikesoundfont le
-
YOUR second and third sample contain that square rumble, but the first, imported and exported three times in my 2.5 version comes out the same, no square wave pollution.
-
bottrop, le -YOUR second and third sample contain that square rumble, but the first, imported and exported three times in my 2.5 version comes out the same, no square wave pollution.Ok, thank you for your attempts of reproducing results.
Then I would like to confirm what you did, because now I want to now know how the square rumble in my data comes in.
I guess what you did is
1. Created new soundfont and import the first sample.
2. Saved the soundfont.
3. Exported the sample once.
4. redo (1~3) procedure 2 more times.
5. Compared the newly exported 3 samples.
because in this case I could reproduce your results. No difference among these 3 samples were observed.
But I still found difference between the original file and the 3 newly exported samples, and square rumble in the 3 samples compared to the original was observed.
I guess exactly the same square rumble was added to the 3 samples, so the 3 samples are identical.
The more detailed procedure I did before create this discussion is.
1. Created new soundfont, imported the original file and saved the soundfont.
2. Closed the polyphone, and opened again.
3. Opeded saved soundfont and exported sample file.
4. Created a empty preset and saved the soundfont.
5. back to procedure 2.
(I do not have so much confidence that the procedure 2 was always conducted. I might skipped some of the time.)
In this case, every sample I exported turned out to be different, as you confirmed in the second and third file's square rumble.
My hypothesis at this point is that when you save, or when you load soundfont(samples), polyphone do some kind of modification on the samples.
But I think I have to examine more in detail to fully identify the cause.
(I am sorry to say that my reply will be more or less late, partly because this phenomena itself is very subtle, hard to explore and need time and concentration.
I really thank you for showing interest my post and trying to reproduce the phenomena.) -
Message de musiclikesoundfont leIt comes to my mind that I have some very basic important information left to inform.
Firstly, I saved soundfont as .sf2, which I believe the default of polyphone.
Secondly, in Viena SoundFont Editor, this phenomena was not observed.
That's why I believe this phenomena is not common in sf2, rather polyphone specific. -
i think it is interaction between your system, soundcard or machine and Polyphone.
i am in Windows10, Soundblaster Audigy and play throug my stereo. -
That's really one possibility.
I am also in Windows10.
I usually used NW-A50 as USB-DAC and played through it.
Do polyphone pass through these sound systems when it just load and save the soundfont?
If it is, it is highly possible that everyone's playback environment could affect the outcome.
Because I believed that polyphone just save the same sample and do not pass the sample to playback system, I really felt weird when I found that the sample was changed.
Now I start thinking that I might misunderstand about that.
Anyway, I will check whether this phenomena is also observed in another PC, and without NW-A50. -
Message de musiclikesoundfont leThe same phenomena was observed when I tried in other PC, of windows11 and different sound system.
In this PC, sound was played through a monitor.
The playback setting of windows was confirmed to be 16bit, 48000Hz.
The first sample in "16bit outcome of the procedure.zip" (16bit, 48000Hz) was used as original.
The same procedure I described above was conducted.
The original file, exported file after the soundfont was saved, closed and opened again, secondly exported file after emply preset was added and the soundfont was overwritten(saved), closed and opened again, these 3 files were different according to foobar2000 bit compare tool.
Now I have really no idea about what is the possible cause of this phenomena.
Because of this, I now do not have any plan to examine.
Anyway, I hope these posts may give some help to those who conflicted similar phenomena.
I will give some polyphone setting screenshot later, although I have not so much confidence that any of these settings could affect the phenomena. -
Message de musiclikesoundfont le
Connectez-vous ou inscrivez-vous pour participer à la discussion.
Polyphone a besoin de vous !
Polyphone est gratuit mais il y a des coûts associés à son site web et à son développement. Un petit coup de pouce aidera beaucoup.
Faire un don
Apprenez les bases
Voir le tutoriel
Haut de
page
page








