I wouldn’t have objected to knowing what the WAV format was. Can we assume you kept opening up the last export?
And nobody would complain about seeing a spectrum of the original “show” as a control.
Is that the Audacity spectrum analyzer?
<<<…but I can heart it clearly…>>>
After you crank the monitor volume all the way up? Can I guess that would not be the setting you would use to listen to your Beastie Boys?
Are we looking at the reason not to do serious production at 44100? The maximum unambiguous (non-noisy) show sound coding at 44100 isn’t 22 KHz, beyond human hearing. It’s 16.9 KHz, right around the point where you show terrific errors increasing.
Koz
When CD Audio was designed, somewhere around the Harding Administration, there was a serious fight over the amount of music on the disk versus the quality. There was no serious compression available in the Harding years.
When they settled on 80 minutes and 44100, all the golden ear audio people rose up as one and complained that they could hear the damage. Now you can, too.
Koz
I’m not sure that you can “hear it clearly”. -76dB is low level and below the noise floor of many domestic amplifiers.
The noise that you can see is caused by dithering. If you create a silent track in Audacity and export it as a 32 bit WAV file there will be no dithering applied and it really will be silent (all samples at zero). Is that noticeably more silent than your 16 bit silent recording? Any noise that you hear from playing the 32 bit silence is from the playback equipment and not from the audio track.
You can eliminate dither noise by exporting as 32 bit.
<<<-76dB is low level and below the noise floor of many domestic amplifiers.>>>
But even if it’s not (my BGWs can do 106dB) you can’t listen to the Beastie Boys and then just press Stop and hear the noise – particularly if it’s up that far in frequency. Most people can’t hear much past 15 KHz, the frequency where glass US TV sets used to sing.
<<>>
I don’t entirely agree. The dithering happens at the sample rate, not the bit depth. That’s how they get the frequency response out to 20 KHz. Add dithering noise so the sampling rate error doesn’t ever line up with the music. We get away with this because 20 KHz distortion energy appears at 60 KHz.
It doesn’t matter how accurate the sample data is if the music arrived between the samples.
Koz
Very good point if listening to a CD player, but not necessarily such a good test when listening to audio playing on a computer. Some sound-cards will shut off all output when there is no audio data (effectively gating the output) thus appearing to have less self noise than they really have. This is common on PC laptops which often have very noisy sound-cards (and when played through a good sound system the gating can be quite obvious).
That’s not correct. Dither is applied when downsampling the bit depth.
That’s why a steep slope low-pass filter below the Nyquist frequency is used. This is known as “anti-aliasing filtering”. High quality digital audio devices have some very fancy ways to improve anti-aliasing filter beyond the limitations of conventional low-pass filters, but the Nyquist limit still holds true and thus for 20/20k audio sample rates of 48kHz are still better than the 44.1 CD standard.
I tried that, no graph was displayed when I tried to analyse the original “true” silence (created in Audacity using “generate silence”),
bear in mind it was a true flatline, zero amplitude.
I think you may have something here: the only conversion I can think is taking place, which could introduce noise, is “32bit float” wavs being saved as 16bit, then converted back into “32bit float” when I reopen it. Perhaps this 32-16-32-16… is what is adding the hiss noise.
Yes. Annoyingly a bit more hiss was added each time I opened it then saved it, (like bad old analog recordings).
I appeciate this accumulation of noise will occur with lossy compression (e.g. MP3) but I though it would not occur with lossless uncompressed wavs
How do I export as 32-bit wav in Audacity ? …
<<<I appeciate this accumulation of noise will occur with lossy compression (e.g. MP3)<>>>
Then you may be appreciating the wrong thing. MP3 damage has nothing to do with hiss and noise. If anything a highly compressed MP3 is quieter than a WAV file. MP3 damage is in quality of performance.
My silly illustration of MP3 damage is the ability to convert two cheap violins and a Stradivarius into three cheap violins. If there is low background hiss and noise in the original performance, it’s likely as not to vanish.
Koz
Conversion in Audacity is not straightforward. All the spells are buried in here.
http://forum.audacityteam.org/viewtopic.php?f=11&t=10190&start=0
Koz
No, but technology may be against you here. Audacity does everything internally at 32.
Koz
Also, the noise density is very different.
Koz
I’ll try keeping tracks as Audacity projects when editing and only convert to wav in the final mix.
Yes there was a slight improvement, but still an audible hiss which grew with each open-save operation.
<<>>
This might be where a little thinking outside of the box is handy. Because the problem is different between the two samples, that might tell us something about how the problem was created, and further suggest a way to get rid of it. We can’t get stuck in the Broken > Fix It > Broken > Fix It loop.
Can you make it worse? I’m not kidding. Is there something you can do that makes the problem much worse?
Koz
Sorry for the delay - my internet connection has been down this evening.
As Koz said, Audacity processes internally at 32 bit. When you Export the audio as a 16 bit file, dither is added. The dither noise is introduced when down-sampling from 32 bit to 16 bit. Importing a 16 bit file or up-sampling do not introduce any dither noise.
You will notice that the noise is above 15kHz - that is because the type of dither you are using is “shaped” to push the noise where it will be least noticeable while still performing its necessary function.
Other types of dither produce different responses, for example “triangular” produces a lower level of noise, but across the full audio spectrum.
Setting the dither to “rectangle” will introduce virtually no noise when down-sampling a silent track, however the dither is likely to be less effective and more noisy on a non-silent track.
“Triangle” or “shaped” dither are usually considered to be best for music. My preference is for shaped.
To avoid the dither noise you can switch dither off in Preferences (not recommended), or Export as 32 bit audio. To Export 32 bit audio you need to select “other uncompressed files” as the Export file type, then in “Options” you can select a 32 bit format (such as 32 bit signed Microsoft WAV).
It’s curious to think that virtually every CD of the last 20 years has dither noise present (though usually not in the silence between tracks).
Thank you Steve, I’d never seen this meun before.
After three export-open operations at 32-bit the silence is still a true flatline. Problem solved. Hoorah ! 