Wavosaur

[Radiowaves]

Hi,

first of all, thank you very much for this great small piece of software! Wavosaur has all the features (and a lot more) that one needs to prepare audio files prior to making MP3 files. And it has a much more professional look-and-feel than Audacity (actually it looks a little bit like my beloved Sound Forge). So I'd like to recommend Wavosaur to everybody who needs a software for this job.

But...

during my first check I found out that some dithering is added already when opening a wave file (16 Bit, 44.1 kHz) and saving the file without any changes. Dithering is on the lowest bit only.

While dithering is great for reducing bit depth at special operations (filtering, ...) I think that there should be added no dither when simple cutting, pasting and fading is applied to sections of the audio file. What I would expect is that the original wave file is unchanged with exception of my intentional changes. If I only add 2 seconds of digital silence or something like this to the file I do not want to have it dithered. Other software (Audition, Sound Forge) is working in this way.

So, is there a way to disable this dither (maybe in the ini file) or to make it switcheable in the audio settings of future releases?


Thank you very much!

[Rex]

Hello,
Welcome to the forum

can you upload somewhere a "before / after " example.

There is no dithering involved in the saving or loading, if the file is different, it is a bug

[Radiowaves]

Please find an example here.

"test" was generated externally and loaded into wavosaur. "test return" is the result of simple saving the original test file again.

It was not as easy as expected to find a "simple" example. I thought that pure digital silence should do the job, but it did not. Digital silence returns digital silence. I had to add some modulation to find the effect again. Please look at the difference between "test" and "test return" amplified by 80 dB. You can (of course very distorted) hear the original signal - given back with sample values of -1, 0 and 1 (or with the sample values of -10.000, 0 and 10.000 when amplified by 80 dB). This also works with white noise and music - the structure of the original can be heard.

This is very strange to me. To avoid calculation errors I used 2 different editors to make the "difference" file - same result. Am I wrong?

[Tron04]

The "bug" can be simulated very easily:

1. In Wavosaur, choose "File/New" with 16bits, 44100 Hz, 20 seconds length.

2. Choose Tools/Synthesis/Frequency Sweep/Linear

3. Press G to create a Sonograph. Notice the clean black areas in the sonograph. This is OK.

4. Save the Sweep to the disk.

5. Close all Windows and load the saved file

6. Press G to create the Sonograph again. You will see the dither that has been added.

Cannot tell where the dither is added and if it stays in the wav when saving. But here you have a procedure to reproduce it.

[Wavosaur Main Developer]

It's not a bug, it is the 16 bit resolution: try to save your wav in 32 bit float and it disappears.

Wavosaur synthesis uses the maximum internal bit depth to have highest precision: when you save in 16 bits, you see then the quantization effect.
_________________
~~~~~~~~~~~~~~~~~~
Wavosaur Main Developer
~~~~~~~~~~~~~~~~~~
http://www.wavosaur.com

[Tron04]

OK, I understand.

Nevertheless I think there should be a preferences option that lets the user decide if dither (and optional noise shaping) is applied when saving.

[FordMustang]

what dither ?

[Radiowaves]

[Tron04]

Even though I understand that the internal precision is 32 bit, I do not like that simply by saving a 16bit source wave again in the same format gives me an output file with dithering that might be unwanted.

Please make dithering when saving an option that can be enabled by the user in the preferences.

[FordMustang]

there is no dither