I have been experimenting recording mono soundtracks at 44100 HZ, 32 bit sound,
and have been saving them to mp3 files successfully. I have observed
it inserts a second, empty sound channel (probably to do with stereo, I guess.
-I want to know how to normalise the digital volume across a number of sound recordings.
All of them have the output volume setting saved to 1.00, but still they
differ in volume at the data level. When I go to the Amplification effect,
I do notice that different numbers come up. Do I just pick an effective, common
value across all my sound files?
It won’t do unless you have panned the track hard left or right using the L…R slider on the Track Control Panel. Don’t pan the track if you want to export to a mono file from a mono track.
The output volume slider in Mixer Toolbar:
does not affect the exported file, only the output volume of the audio device selected in Device Toolbar .
It’s probably better to use Effect > Normalize… (you would definitely want to do this if you are using Export Multiple… to export multiple files at one go). Just choose the value you want in Normalize for how loud the loudest peak in the file should be. This says nothing about how loud the files sound, it is just setting the peak value.
You don’t have to import them at the same time, but you want both of them in the same project window. To achieve that, use File > Import > Audio… and select the file or files you want. If the files you want are in the same folder, you can click on one file to select it, then hold CTRL on your mouse then click on the other file, which selects both files.
Alternatively, drag the files into the same project window.
As above, you want both files in the same project window.
As with all effects, only the selected track(s) are acted upon. So you need to select both tracks to normalize both of them. If you only have two tracks then CTRL + A or Edit > Select > All will select both of them. Then with both tracks selected, and without any tracks playing, choose Effect > Normalize… .
Leave L and R as they are in the centre for both tracks. When you have normalized them, use File > Export Multiple… then both files will be exported separately (in mono) one after the other.
You should be able to export mono tracks without an empty second channel.
Ensure that you do only export the selected track or that all other tracks are muted and that it isn’t panned to any side.
Also, the track must be mono in the first place. Split it to mono if there are two wave forms and delete the empty track.
This will almost double the quality for a given bit rate.
I am not sure what you mean with differing volume at data level.
Do you mean that the re-imported files have no longer a amplitude of -1.0 dB?
That’s unfortunately a normal behaviour for mp3 conversion.
It always depends on the content, how much the volume will increase (or sometimes decrease).
If you export simple white noise at -6 dB, the amplitude will go up to 0 dBBS after conversion (at 128 kbps).
You can simulate this effect by employing a low-pass filter around 20000 Hz, >= 12 dB.
A generated sine wave will be untouched whereas a square wave has afterwards a peak 3 to 4 dB higher.
You have to explain what you’re measuring exactly if you do not mean this volume differences, i.e. for the whole track, after re-import.
Firstly, export the file at a level of about -4 dB.
The reasons for this are stated above.
Then, use an external program to adjust the level according to the Replaygain algorithm.
I personally use Foobar2000. You can analyse whole playlist by simply choosing one of the Replaygain options from the context menu.
The corrected value can either be applied permanently or as additional information in the metadata. Some players can read this tag and set the volume accordingly.
-What can I do to eliminate such a difference, and normalise
both their volumes relative to one another, at the data level?
[u]ReplayGain[/u] (or the related MP3Gain or WAVGain) will automatically adjust the perceived volume.
Perceived volume is related to the average level and the frequency content. The Amplify effect and normalization target the peak levels which don’t correlate will with loudness.
ReplayGain works along with your player software to adjust the volume at playback time. It adds a “tag” to the file, but doesn’t touch the audio data. Of course, your audio player has to support ReplayGain. iTunes and the iPod have something similar called Sound Check.
WAVGain adjusts the actual volume of the file so no special player software/hardware is required.
MP3Gain also adjusts the actual volume in the file. Due to the way MP3 in encoded, the volume can only be adjusted in 1.4dB steps (without re-encoding the MP3), so it’s not quite as precise. (If you have the uncompressed original, you can use WAVGain before compressing to MP3.)
Note that most digital files are already normalized. This means that in order to match the volume of two or more files, you generally have to reduce the loud file rather than boost the quiet-sounding file. If you ReplayGain all of your music, you’ll notice that most of your songs will be quieter. You can boost the target volume for MP3Gain, but that means you are giving it “less room to work” and you’ll have to choose between doing nothing or allowing clipping (distortion) with many files.
If you have only two songs (or a few songs), it’s possible to match the loudness with WAVGain and then adjust all of the songs by the same amount so that the song with the highest peak hits 0dBS. That’s more work, but you’ll end-up with the loudest possible matched volume.
Actually I do not know what Zachary1234 meant by setting the output volume, I thought he meant the output volume slider given he said “1.00”.
For what it’s worth, I only see the volume go up to -1.6 dB in the exported 128 kbps MP3.
For a normal song the volume change will not be anything like as large. The problem happens because MP3 is a lossy format that throws some data away so as to reduce the file size. Audacity does not make compensatory changes before or after encoding to try to make the outcome normalization level the same as that requested. A few people think Audacity should do that.
Even if you don’t want to use ReplayGain you should not export an MP3 normalised to 0 dB (the highest peak level possible) because the exported volume could then go above 0 dB. If that happened it could sound a little distorted. If that happens, you can fix it by reimporting the MP3 (as long as you import the file as 32-bit float which is default) by going to Effect > Amplify… , choose a lower New Peak Amplitude then export again.
If you do not want volume changes on export, then you should use a lossless format like WAV (but the exported files will be much larger).
Sorry Gale, I meant -4 dB white noise which will go up to 0 dB.
The overshoot is essentially increasing the louder the audio is, i.e. 3 dB louder isn’t equal to 3 dB. This might be due to the uneven mp3-quantization. Also, VBR and CBR do not exhibit the same increase.
It is evident that a -4 dB gain and subsequent replay-gain adjustment is the safest way.