My 1-Minute Test - USACoder - Joe McVerry [OK]

[Usacoder]

The link: https://librivox.org/uploads/tests/test_usacoder.mp3
Software Audacity 3.1.3
On Ubuntu
Using an analog microphone.

[Winnifred]

The link: https://librivox.org/uploads/tests/test_usacoder.mp3
Software Audacity 3.1.3
On Ubuntu
Using an analog microphone.

Welcome to Librivox, Joe!

Your voice is clear and animated. However, your test has a couple of issues that need to be addressed. The volume is too high, at 92.8 (acceptable Librivox volume range is 86 - 92 dB, with the ideal being 89 - 90 dB). You can fix that by selecting the whole track, and using Loudness Normalization, set to -19LUFS, under Effects to bring the volume down to an acceptable level.

However, you also have a high amount of DC Bias. DC Bias occurs when the shape of an audio waveform is not centered, so that the height of the empty space above the top of the waveform is not the same as the height of the empty space below the waveform. if you look at your file in Audacity, you'll see the wavy line going well below the centre line, especially where it drops drastically at the beginning. The Checker is telling me that your DC Bias is at -1009 (-3.0788%). Most recordings are at or near zero. I'm not sure I know what causes this (I've never seen one this large), but when I Google it, I find this explanation: "a fixed voltage offset somewhere in the audio chain before the analog signal is converted to digital values." Although you could try to repair it using Normalization under Effects, that won't fix it completely. Perhaps try to record again, and see if it keeps happening?

It's good practice to remove any DC bias as a first step when you start editing. Although DC bias is not a problem on its own, it can become a problem during editing. For example, audio with a DC bias becomes clipped more readily than audio without.

You might also try asking a question over in the "Need Help? Got Advice? forum. A member with experience with DC Bias issues and Linux might be able to offer more tips than I can.

If you re-record and post a new test here, I'll take a look at it to see if it's still happening.

Thanks,

[Usacoder]

Winnifred,

Thank you for your analysis of my recording.

I reposted an update at https://librivox.org/uploads/tests/test_usacoder.mp3

It's the same recording. When I made it the first time with Audacity I did no editing on the recording. For this version, I edited the original file and used three "Effect" options:
1) applied the "Loudness Normalization" option.
2) "Noise Reduction"
2.1) selected the last 3 seconds of the recording, the no talking part, to get a "Noise Profile".
2.2) selected the whole recording and applied the "Noise Reduction" filter. I made no changes to the options for this step.
3) applied the "Normalize" option.

The recording was saved and uploaded.

Regards,
Joe.

[Usacoder]

Winnifred,

I just found the Checker program that can be used to validate MP3 files for LibriVox.

After running the updated test file through it, Checker reported that the volume was still too high.

So I edited the file with Audacity and applied the "Loudness Normalization" effect a second time. The Checker program liked what I did.

I reuploaded my test file.

Regards,
Joe.

[Winnifred]

Winnifred,

I just found the Checker program that can be used to validate MP3 files for LibriVox.

After running the updated test file through it, Checker reported that the volume was still too high.

So I edited the file with Audacity and applied the "Loudness Normalization" effect a second time. The Checker program liked what I did.

I reuploaded my test file.

Regards,
Joe.

That sounds good, but I don’t see a link here. Can you post that too, please? I don’t have access to all the files, just the ones that are posted here.

Thanks,

[Usacoder]

The uploader uses the same file name: https://librivox.org/uploads/tests/test_usacoder.mp3

[Winnifred]

The uploader uses the same file name: https://librivox.org/uploads/tests/test_usacoder.mp3

Ah, right. I'd already started a response to you on the off chance that you'd reposted with the same url. I checked the previous file, and it's all good now.

I think the volume change might have been because you used Normalization last, and that brought it back up again. You might want to play around with the input gain on Audacity before you start recording, to reduce the need for loudness normalization. You want to be careful not to apply too much editing, as it might start to distort your voice.

To set volume in the Audacity Gain Slider and on your mic, see the top of the Audacity screen



Move the input gain slider (upper row beside the little mic icon) to whatever spot yields the best volume level.

Then adjust the input gain dial on your mic, if you have one, so your raw recording has wave forms that peak at about the 0.5 marks in the Audacity window, like this:



There's also another option for double-checking volume in Audacity. Download the free ReplayGain plugin for Audacity to find out exactly how far your recording is from 89 dB. If the level is at or near zero, your volume is very close to 89 dB. If between +3 and -3, you're within the acceptable range for LibriVox. Instructions for installing/using this plugin are here. (Assuming this is available for Linux, but it's been around for awhile, so I suspect it will be.)

You're ready to start recording! I suggest heading over to the "Readers Wanted: Short Works (Poetry & Prose)" forum to find your first recording. There are lots of short options there in all genres.

Happy reading!

Cheers,