*why is there still noise even when...

[npags711]

...you silence the audio = Ctrl + L

Even after I silence certain sections of audio in Audacity, and export the file, when I play it back on various media players - there still exists a static noise...

The last project I worked on - I noticed this and cut the static noise with Adobe Soundbooth - then exported via Soundbooth and the noise was eliminated.

So why, even after silencing the audio, does static noise still exist after export out of Audacity?

And is there any way to prevent this?

Thanks!

[steve]

What level of noise are you referring to? (in dB relative to full scale)

[DVDdoug]

If it's very-very low-level noise that you can't hear without cranking the volume way-up, it might be dither.

Dither is added noise, but it's supposed to improve the sound. Under "normal circumstances", the dither is over 90dB down, and you can't tell any difference with or without dither.

I believe Soundbooth also has a dither option.

[npags711]

Sorry...mean to reply back way, way sooner...

The dB = -70dB to -72dB

You guys can silence a given section of audio and export - then listen to test and you'll prob. get the same thing...

I just don't get why sound still exists even after silencing in Audacity...and why SoundBooth's noise removal took care of it without issue

Let me know when you catch a minute - I'll get back promptly from here on out.

Thanks for the help - I really appreciate it!

[steve]

That will be "dither noise" and is because, by default Audacity processes in very high quality 32-bit float format, but you are Exporting in a lower quality format (probably 16-bit). 16-bit audio is not able to reproduce audio below about -80 dB without introducing severe distortion.

In terms of absolute audio fidelity, the best option is to export in 32-bit float format. This will export the audio exactly as it is in Audacity with no quality loss and no added noise. Unfortunately few other audio programs or hardware support 32-bit float format.
The next best is to use 32-bit (integer), but that has the same problem of few other programs or hardware supporting the format.
The next best is to use 24-bit (integer), but that has the same problem of few other programs or hardware supporting the format.
The next best is to use 16-bit (integer) and apply "dither". This format is supported by just about everything. The downside is that it introduces a small amount of noise (in the region of -70 dB peak).
The next best is to use 16-bit integer without "dither", but although this does not add any dither noise, the format cannot reproduce very low level audio without introducing severe distortion.

The trade-off with using dither is noise vs distortion, and the noise is generally thought to be less obnoxious than the distortion, except for times when there is supposed to be absolute silence. If you are making audio CDs and you want a couple of seconds of absolute silence between tracks, trim the tracks down to remove silence from the end of the track, then use the CD burning software to add a couple of seconds of silence (most CD burning programs will do this by default as it is the Red Book standard).

As a little demonstration - here is a -80 dB sine tone (440 Hz) that has been converted to 16-bit format. First with "shaped dither" and then without dither.
Which do you think sounds more like a pure sine tone?

[npags711]

Thanks Steve, I really appreciate the info and detail you provided...you the man!

Regarding your attachment, the second one - without dither is definitely clearer with a sharper sound.

So if I was exporting in 32-bit = the best option would be to select "other uncompressed files" as file type and select the 32-bit float?

I'll give you an idea of what I usually do...

For most of the audio I edit, the final export is an MP3 - 128kbps.

I do editing in Audacity and then export an MP3 at 128kbps and import that file into Premiere Pro - where I add intro/extro music, cut into tracks/cd's, and do some final sweetening. Then export thru Premiere Pro by MP3 format at 128kbps.

Obviously, this isn't working b/c I want to get rid of the "dither" b/c it's like background noise that's not needed. I want it exactly like I've edited in Audacity.

Am I able to export in 32-bit float considering the process I usually use?

What do you suggest is best to export out of Audacity without "dither"/exactly as I've edited in Audacity, considering that the final export will be an MP3 @ 128kbps?

Thanks Steve - let me know when you catch a minute

[steve]

I do editing in Audacity and then export an MP3 at 128kbps and import that file into Premiere Pro - where I add intro/extro music, cut into tracks/cd's, and do some final sweetening. Then export thru Premiere Pro by MP3 format at 128kbps.

MP3 is a "lossy" format. That means that when the audio is encoded (as MP3) some of the audio information is thrown away. At 128 kbps, the sound quality loss is noticeable if you compare the MP3 with the original on a reasonable sound system (for example, cymble sounds tend to become "zingy").
If you re-encode an MP3 (as you are doing in Premier Pro), then a bit more of the audio information is thrown away, deteriorating the sound quality further. To avoid this, the Export from Audacity should be in WAV format so as to avoid throwing away audio information twice. Encoding as MP3 should be the last step in production and only done once.

As 128kbps is a relatively low quality audio format, I would not expect that dither noise would make it significantly worse. I would Export from Audacity as 16-bit WAV (with dither), then use that in Premier Pro.

the second one - without dither is definitely clearer with a sharper sound.

Yes it's "clearer" and has less noise, but it sounds a lot less like a 440Hz sine tone. I should have included an example of what it is supposed to sound like along with these two samples.
Here's three samples for you to listen to (there is a 0 dB "blip" between each sample which will assist in setting the playback to a realistic level).

[npags711]

Steve - thanks again...

I heard previous to submitting the post regarding the process I use...that you shouldn't export MP3's twice...should have known that with my video experience but thanks for the reasons provided.

The dither actually is pretty noticeable - I do want to export exactly as I've edited in Audacity and at the highest quality possible before final editing and export to MP3...as mentioned.

So, one last thing...

How do you suggest I do so? Go with 32-bit float and select "other uncompressed files" or whatever that option is...?

And regarding the dither sample you provided in your last post - I figured the answer I gave wasn't exactly addressing the point you were trying to make. Thanks for providing the other sample

Cool buddy - thanks!

[steve]

The dither actually is pretty noticeable

Do you think it's "pretty noticeable" on the last sample that I posted? (440hz-80dB.wav)
(different solutions depending on your answer)

[npags711]

Steve - the samples you provided...I can hear faintly the background noise...it's not that noticeable

My main aim is to have the highest quality possible and when I silence the audio sections in editing and remove background noise I want my final product to precisely match that which I edited - same quality.

So I don't want any dither - or anything added or different from what I edited.

So what do we think the best export option is?

Thanks Steve