I have been using Audacity to convert LPs to mp3 tracks. I generally will record one side and then apply some noise reduction. Then I use the “generate silence” function to place absolute silence between the tracks. Following this I export each selected track as an mp3, usually 128 bit. The generated silence in truly silent as far as I can tell. I use headphones to monitor the sound. But when I import the created mp3s back into Audacity, I find that the silent portions have a slight amount of low level hiss. Is this low level hiss a normal part of the encoding process or is there a way to prevent this from happening?
Not sure why you are adding silence (with the Generate Silence function) to the inter-track gaps.
I transcribe a lot of LPs what I do to the intertrack gaps after capture and cleanup is to fade out the end if the prior track (if needed), dade in the succeddding track (if needed) - then select the intertrack gap and silence it with the CTRL-L shortcut (the k/b shortcut for Edit>Silence )
I export as WAV from Audacity, burn CDs and convert to AAC in iTunes - with no hiss problems in the intertrack silences.
WC
Exporting silence as MP3 may introduce a very very low level of hiss, but it should be virtually inaudible (around -70dB or less). It is more likely that the background hiss that you can hear is from your amplifier or sound card.
As WC says, I’m not sure why you need to generate silence for this. If you have several audio clips with gaps between them, the Exported audio will replace the gaps with silence (or very very close to silence - Exporting in WAV format and the silence will be absolute as long as you are not changing the bit depth).
Thank you for your reply. OK, I understand your methods for transcribing and creating MP3s of individual tracks. I know it isn’t my amplifier because I do get absolute silence without hiss when working with the first part of the digitization process. It seems that the noise is created during the MP3 encoding process. Remember, I am using high quality headphone to monitor the process. The noise level is very low and is not noticeable through my speakers. I used to use the GoldWave software for digitizing my LPs, but I prefer Audacity for its functionality and simplicity. However GoldWave is able to export silent data to MP3 without any hiss.
What level of hiss are you talking about? If you import the track back into Audacity and zoom in on the vertical scale (click on the vertical scale) do you have a noticeably wiggly line in places where it should be silent?
Does this only happen with MP3 export? If you Export as WAV, are the parts that are supposed to be silent actually silent?
I saved it as a .wav and it also produced the low level hiss. I amplified this hiss and it appeared as random triangular spikes, both positive and negative, so it probably has nothing to do with mp3 encoding. But while experimenting with the preference settings, I discovered what was causing the problem. In Audacity Preferences under the “Quality” settings, I noticed that the default setting for high quality conversion was “Fast Sinc Interpolation” and the dither was set to “Triangle.” I changed the converter to “High-quality Sinc Interpolation” and the dither to “None”, and that solved the problem.
I just went back to recheck what I just did and it seems that it’s the dither setting that is the factor for the hiss, so setting dither to “None” is all I needed to do.
That is not normally recommended - “dither” is designed to improve the sound quality when reducing the bit depth.
What format are you exporting as?
“Shaped” dithering will usually be less noticeable than other options, though switching to “rectangle” dither should remove the issue with dither being applied to silence.
See here for information about dither: http://en.wikipedia.org/wiki/Dither
For best quality you should set the default bit depth to 32bit (float). For high quality audio the exported format should not be below 16 bit. MP3 files will always be lower quality than WAV - 128kbps is usually considered to be the lowest (most compressed) rate for reasonable quality music files.
Steve, thanks for all your help. I think I have a better understanding the purpose of dither in the processing of the data. I also agree that using rectangular or shaped dither produces the results that I was looking for.