Hey guys. I’ve just recorded and edited an audiobook version of my book. In order to preserve the freshness of my voice, I recorded only two chapters a day. That meant that, although I tried to place the mic exactly the same way each time, and have the mic level consistent, my levels from chapter to chapter vary a bit. Will using the “normalize” effect on each chapter take care of this?
Also, what kind of file should I export the tracks as – that would be best for audiobooks/CD’s?
Is this a Windows machine? How did you have the microphone connected and what kind is it? If you notice changes to your ear from one session to another, that’s actually a fairly major volume shift. Windows machines like to “help you” and assume you’re a business executive on Skype. If you leave the speakers running while you talk, for example, the computer may do a number of very odd things to your voice.
To specifically answer the question; maybe. All Normalize and Amplify do is change the volume of the whole show as a single group so the one loudest part changes to a specific value. It may help, but most people want those tools to manage the volume minute by minute and they won’t do that. For example, if the one loudest part of one chapter happens to be you dropping a pencil, then that pencil takes over the whole show.
Try Amplify with the default settings. If you were careful with your recordings, that may help – and that may be enough.
But the recordings should not be doing that in the first place, and we recommend against trying to fix poor computer adjustments in post production audio tools. Do you wear headphones while you record? That can go a very long way to solving sound level problems.
If you’re talking about a fairly small amount of variation, then that is to be expected. If you’re talking about a big difference, then there’s a problem.
Assuming that you mean that it’s a fairly small difference:
“Will using the “normalize” effect on each chapter take care of this?”
The problem is that “normalize” (both the Normalize and Amplify effects) work only on the level of the maximum peak in the recording. If the recording on one day has one super loud peak anywhere in the track, then normalizing will suppress the rest of the track to a lower level. There are also tonal differences (timbre) to consider. If your voice was a little more stressed on one day, then the recording may sound a bit louder, or quieter, than the recording from another day, even if the levels are identical.
Recording an audio book is a lot of work and I would not recommend taking short-cuts here. The only way to match the levels from one day to another properly is by listening and manual adjustment.
After you have recorded all of the chapters, check through them and find the quietest one. That will be your reference level.
Use the Amplify effect on all of the other recordings to bring them down so that they sound right compared to this reference track.
That is an excellent tool if the goal is for a “radio voice”. For evening out the dynamics it compares extremely well against any similar device at any price (and it’s free).
For automating the process of “levelling” the dynamics, compressors are the tool for the job (hence their use in live radio) and Chris’s dynamic compressor is probably the best one for this Audacity job. However, if the aim is to keep the natural dynamics, then it must be used very carefully as it can easily be overdone.
My working method would probably be to use both manual adjustment, and subtle use of a compressor, though the order in which they are used will make a difference. For best results it may be necessary to adjust manually, then use the compressor, then minor adjustments made manually. I would highly recommend experimenting on some short samples before doing it for real, and definitely keep safe backup copies of your original WAV files.
If you are going to use any sort of Noise Removal, do that before using the compressor.
I recorded the audiobook with a Plantronic headset. The headset’s recording volume was consistent; but the exact placement varied slightly when I would take of the headset and put it back on, from chapter to chapter [even though I tried to keep the mic in the same position.]
Thanks for the recommendation, Trebor. I’ll check it out.
Thanks, Steve. I’ll try to manually adjust the levels from track to track. Since I’m a novice, I may just leave the natural voice, rather than the compressed for now.
An excellent idea, however, you have arithmetic working against you. A one inch shift in placement of the microphone can double the distance to your lips, and yes, that can have significant effect in the show.
O.K. So I just stacked all the tracks into one file, then manually adjusted the “Amplify” so that levels were consistent from track to track.
Now, what kind of file should I save each track as? I’m going to burn all the tracks onto about 4 CD’s and ship them to the duplicator.
44100Hz, 16 bit (Microsoft) PCM WAV
For the 44100Hz, part, look in the lower left corner of the main Audacity window for the “Project Rate” setting and check that it says 44100. If it is at any other number, change it to 44100 before you export.
For the other settings, select “16 bit (Microsoft) PCM WAV” in the Export dialogue (assuming you are using Audacity 1.3.12).
44100, 16-bit, Stereo is the format of the sound on a Music CD. It’s an uncompressed, non-damaged format. It’s handy to save your work like that (on multiple hard drives) as archive and safety backup.
From those files,you can go downhill and create MP3 or AAC or other portable formats, but you can’t come back up from MP3 once you do that. It’s a one way trip and MP3 damage is permanent. So carefully archive your WAV files.
You might also make use of mp3Gain, which is a free app that takes a group of mp3s and - through some fancy-smancy algorithm - makes them “sound the same to the human ear” etc. I’ve done just some basic tests with my own audio book segments, and - while subtle - it could prove a useful tool.
And one other thing as just an insight from someone who listens to a lot of audio books: so long as you have consistency within the chapter, the ear is pretty forgiving. Even noticable differences fade into the unconcious after a few seconds, and you’re right back “in the action” so to speak.
Any moment of noticeable inconsistency within the chapter, however, are very distracting and ruin immersion. I’m dealing with this when trying to do some punch in corrections on a different day than when the recording was first done. It’s pretty much unusable.
Hope that helps.