https://librivox.org/uploads/tests/test_freso.mp3
Test recorded via my SteelSeries 840 headset mic in Tenacity on (Arch) Linux. I recognise that I need to turn the mic up a bit for future recording. There’s also some background/ambient noise; usually this gets filtered out via a noise suppression filter (RNNoise‐based), but this recording was specifically made of the raw, unfiltered microphone input.
I’m also hoping to be able to get a better microphone this year, but I’m not exactly flush with cash, so I have no idea what the timeline for that will look like.
test - Freso [OK]
Welcome Fresh
I reviewed your test sample, and while it sounds good there are some technical changes needed. LibriVox requires audio that is 128kbps constant bit rate MP3, 44.1kHz sample rate, mono, and between 86db and 92db, preferably 89-90. Your test audio volume is a bit too low at 85.3db. There is also a loud plosive sound on the word "test" when your breath strikes the mic, which is always a problem with headsets.
Make sure the the input gain slider in Audacity (upper right with the little mic icon) is well to the right. Then adjust the input gain on your mic so your raw recording has wave forms that peak at about the 0.5 marks in the Audacity window:
If necessary, adjust the volume level before exporting by going to Effects>Amplify.
Reposition your mic so that it is off-center about 45 degrees. Add a foam wind screen if necessary.
You can pre-check your audio yourself by downloading the free Checker app: https://cgjennings.ca/projects/checker/
Passed with Warnings is a fail with regard to volume.
Please make a new test recording and post it in a reply to this message. I will come back and review it for you.
I reviewed your test sample, and while it sounds good there are some technical changes needed. LibriVox requires audio that is 128kbps constant bit rate MP3, 44.1kHz sample rate, mono, and between 86db and 92db, preferably 89-90. Your test audio volume is a bit too low at 85.3db. There is also a loud plosive sound on the word "test" when your breath strikes the mic, which is always a problem with headsets.
Make sure the the input gain slider in Audacity (upper right with the little mic icon) is well to the right. Then adjust the input gain on your mic so your raw recording has wave forms that peak at about the 0.5 marks in the Audacity window:
If necessary, adjust the volume level before exporting by going to Effects>Amplify.
Reposition your mic so that it is off-center about 45 degrees. Add a foam wind screen if necessary.
You can pre-check your audio yourself by downloading the free Checker app: https://cgjennings.ca/projects/checker/
Passed with Warnings is a fail with regard to volume.
Please make a new test recording and post it in a reply to this message. I will come back and review it for you.
Hi sjmorky 👋
Other than audio being slightly low (which I already acknowledged), which of these technical changes are needed? As far as I can tell the audio I uploaded is a mono-channel 128kbps CBR @ 44.1kHz one, so if my various softwares are lying to me about this, I need to figure out what’s going on with that.sjmarky wrote: ↑January 4th, 2024, 10:46 am I reviewed your test sample, and […] there are some technical changes needed. LibriVox requires audio that is 128kbps constant bit rate MP3, 44.1kHz sample rate, mono, and between 86db and 92db, preferably 89-90. Your test audio volume is a bit too low at 85.3db.
If a windscreen turns out to be a necessity, then I guess I won’t be contributing anytime soon. 🫤
Indeed! I even went so far as to package it for my Linux distribution some days ago, as part of my preparation for making this test recording: https://aur.archlinux.org/packages/librivox-checker 🤭
Besides upping my system input volume I also amplified the audio in Tenacity post-recording and cleaned noise somewhat using Tenacity’s built-in "Noise Reduction" effect.
test_freso_4c.mp3
Your audio meets LibriVox requirements, so you are good to go.
Happy narrating!
Happy narrating!