Post by Yoris on Jan 10, 2019 1:54:43 GMT -7
LibriVox (librivox.org) is a place where you can volunteer to narrate public domain audiobooks (for everyone to use for free as if the recordings themselves were public domain). You can also download and listen to audiobooks that others have recorded.
Useful links:
* wiki.librivox.org/index.php/1-Minute_Test (You need to do this test and get approved before you can start recording audiobooks for LibriVox.)
* forum.librivox.org/viewforum.php?f=21 (You can post your 1-Minute tests here.)
* cgjennings.ca/checker.html (This is the checker software to see if your files pass the technical specifications tests; yes, the checker is available for Linux, too; yes, you can use this on your 1-Minute Test MP3 to help you see if it's in order; however, someone still has to approve your test.)
* wiki.librivox.org/index.php/Amplifying#3.1_Normalize (volume should be between 86 and 92 dB according to the checker; the wiki says ~89 dB)
You may find that your files have a sample rate different from the desired 44,100hz (mine with my digital voice recorder are 48,000hz). To change that without changing the pitch of your voice, import your file into Audacity. Then look at the bar at the bottom of the screen. There should be an option to change the sample rate there (in the lower left of the screen). Once you change it then after you export to a 128kbps constant bitrate mono MP3, the MP3 should be 44,100hz. If you change the sample rate through the thing higher up on the left side, it will change the pitch of your voice.
The digital voice recorder I use can be found here: www.amazon.com/gp/product/B07KBWN8L1/ (it's the 16GB black version). I don't particularly recommend this device for LibriVox recordings, but it does seem to be quite functional for the purpose if you're willing to do some adjustments with Audacity afterward. (I record in the highest quality WAV, edit in Audacity, and then export to MP3 when I'm done.)
For noise reduction, in Audacity, highlight a portion of your recording where you're not reading; then, you can go to Effect->Noise Reduction->Get Noise Profile. Then select the whole recording; then go to Effect->Noise Reduction and click okay. These are the settings I use for my files from my digital voice recorder (your ideal settings may vary):
Noise reduction (dB): 8 (the default of 12 made mine sound synthetic a bit)
Sensitivity: 6.00 (this was the default for noise reduction)
Frequency smoothing (bands): 3 (this was the default for noise reduction)
Sensitivity: 6.00 (this was the default for noise reduction)
Frequency smoothing (bands): 3 (this was the default for noise reduction)
Using the built-in microphone, I don't seem to need to amplify the sound.
Also, you may find that the LibriVox wiki says you need your sample size to be 16-bit. I'm not sure what they mean by that, but it sounds like they want the WAV files to have the format of 16-bit PCM instead of 32-bit float. I don't know if that matters for the exported MP3 or not. Anyway, Audacity probably defaults to 32-bit float and says that no matter what the format of your original file was. So, you can change the setting in Audacity once and it should stay that way. Open Audacity, go to Edit->Preferences->Quality->Default Sample Format. I'm not sure if the checker checks the sample size.
After putting the checking program where you want in Linux, you can run this command to put it in your path (change PATH to the actual path to your checker program):
sudo update-alternatives --install /usr/bin/checker checker PATH 0