How do I set Audacity to meet exact requirements?

Hello There, I am trying to record a sample for Librivox, an online Audiobook site
some of you may be familiar with. However, I just can’t do it. Here are the requirements:

Sample rate: 44.1 KHz
Sample format: 16 bit
Mono channel
MP3: Constant bit rate, 128 kbps
Volume: approx. 88-90 dB, approx. -20 LUFS/RMS
Noise floor: If possible around -70 dB or better.

I have managed to change to Mono - but that’s it! No matter what I do I am unable
to record loud enough - I’m at least 10 dB too quite whatever I try to do.

I am using Audacity 2.0.5 on Windows Vista Home Premium, recording through
a Logitech 340 with USB connector.

What I really need is a step by step Idiot Guide (me being the idiot in question),
but any help would be greatly appreciated! :smiley:

Sample rate: 44.1 KHz

Edit → Preferences → Quality

Sample format: 16 bit

You can ignore that. MP3 doesn’t have a fixed bit-depth. Your headset probably records at 16-bits, and Audacity uses 32-bit floating point internally.

MP3: Constant bit rate, 128 kbps

After installing the optional LAME recorder, you can adjust the MP3 settings when you save (export).

Volume: approx. 88-90 dB

That’s an acoustic (SPL) loudness reference. i.e. It’s the loudness of the sound hitting your ears and it depends on your speakers, your volume control, etc. You may be able to ignore it, or you may need it when setting your LUFS/RMS dB level.

approx. -20 LUFS/RMS

You may need a 3rd-party tool or plug-in to measure that.

Noise floor: If possible around -70 dB or better.

That’s not an easy spec to meet with a home recording. It depends on the noise in your microphone/headset electronics and the acoustic noise level in your “studio”. You may be able to improve it with Noise Reduction or a Noisegate, but these tools can sometimes have side effects, so you have to experiment.

Of course, if you adjust the volume up or down the background noise also goes up and down by the same amount, so you need a good signal for a strong signal-to-noise ratio.

No matter what I do I am unable to record loud enough - I’m at least 10 dB too quite whatever I try to do.

How are you measuring that? (You can easily check the peak level by running the Amplify effect, but that doesn’t correlate well with average “loudness” or RMS.)

You can adjust the volume after recording with the Amplify effect. But,if you amplify too much to the point where the peaks go above 0dBFS, you’ll get clipping (distorted flat-top waves).

The ratio (difference*) between the peak and average/RMS is constant (if you amplify linearly without clipping). So, if you boost the peaks by +3dB, the average will also be boosted by +3dB. So… If your peaks are hitting 0dBFS and your RMS level is too low, you’ve also hit the maximum RMS limit for your particular recording.

If you need to boost the RMS level without boosting/clipping the peaks, you can use dynamic compression (Compressor effect). There are various settings to experiment with, and there are various compressor plug-ins so this may take some practice and experimentation. (Almost all commercial recordings have compression for a consistently strong-loud sound.)

…recording through a Logitech 340 with USB connector.

You may not get “broadcast quality” from a gaming/communications microphone. You’ll get better results with a “studio style” USB condenser microphone (AKA a “podcast” mic). For example, the [u]Blue Yetti[/u] is very popular and has some nice features. (I don’t own one.)



\

  • Decibels are logarithmic so subtraction gives you a ratio.

You can measure the rms value with the contrast tool (Analyze menu)
Librivox has additionally a analyse tool of its own to run on the exported mp3.

The two killer specifications are noise when you stop talking and loudness or density of the performance. In the case of noise, there are published specifications and processes to measure it. Loudness is a magical never-never land. Further, many times if you boost the volume and density of your performance, the noise is likely to rise. So it’s a tennis match between the two values.

I have never produced a -70 noise show in my life and I have serious doubts about producing a -60 noise performance. That’s the kind of result you get in a “real” studio with soundproofing along the walls and your ears pop when you walk in and close the door. -70 is the noise your shirt makes when you settle in your chair or reach for your coffee. This is not a value likely to be achieved by somebody recording in the vestibule of their townhouse.

More normal is -40s and -50s and that’s if you do everything perfectly correct, you have no refrigerator, and you live on a dirt road in Appalachia. Oh and many USB microphones have worse noise than that just when you turn them on. A very common problem is “What’s that thin whiney sound between my words?”

Many posters arrive with the complaint that by the time they got Noise Removal to take all the junk out, the rest of the show sounded so gargly and honky that the client rejected it. Yes, that’s a problem with Noise Removal.

We’re doing a step-by-step for AudioBook Processing.

Do you have any way to post relatively large segments of sound track? Like DropBox or other service? Post 10 seconds or 20 seconds of raw WAV (not MP3) performance and include Room Tone — three to five seconds where you hold your breath and don’t move.

Koz

Many Thanks for all your replies - I definitely have a lot to play with here!

The Folks at Librivox think the volume ‘problem’* is more likely to be corrected
on my PC itself, but the OS doesn’t appear to allow me any access to sound settings
other than the most primitive, like On or Off.

I’ll certainly experiment with all the options suggested, and Kos, I’ll look into some
way of posting a sample when I can.

*The volume isn’t a problem for me, I can hear it just fine on my system - it’s just that the Recorded
Volume isn’t loud enough for Librivox to use. Still, I’ll keep on keepin’ on!

The Audiobook People have to worry about consistency book to book, chapter to chapter and page to page. It’s a similar problem that radio stations have. Every notice when you change stations in your car you typically don’t have to adjust volume? That’s not accidental and lots of money goes into making that happen.

The volume on your music system depends on about a million different settings, so some method must be found to “nail down” performance characteristics so a show appears the same on everybody’s player.

Nobody’s interested in hiking in the forest with a personal music player and having to stop every so often to adjust the sound. I have to do that along noisy streets and roadways and it’s really annoying.

Whether you can hear it or not is irrelevant. It’s not recommended, but I can make an entire sound recording with the speakers turned off just by watching my instruments. That’s the kind of thing they’re talking about.

But sometimes the instruments can be a little magic, too, and that’s the forum message threads having to do with audiobook authoring.

Koz

Well here’s some good news! I seem to have lucked, by what means I know not, onto the correct settings!

Also, some very cool and useful Audacity Plug-ins have been pointed out to me, which will hopefully help
me maintain a consistent quality. Replay Gain is particularly handy.

Anyway, now that I’ve been ticked ‘OK’ I can get stuck into all the Tutorial videos posted on Librivox.
(I’m not planning to become an Audio expert - just competent enough to read some stories that
are pleasant to listen to. :wink: )

The more I learn about Audacity the more impressed I am - looking forward to having some fun with it!

If you’re referring to this plug-in ReplayGain plug-in please note that it is “experimental”.
User feedback regarding success/failure with that plug-in is very useful (also applies to user feedback regarding all plug-ins). Feel free to post feedback in the topic where the plug-in is posted.