Increase volume of speech and reduce volume of applause

Using Windows Vista, Audacity 2.1.2

I recorded an audio of a comedian from an old reel to reel tape into Audacity, but the maximum recording levels achievable did not exceed 12.
On playing back, the voice was low in volume and ‘muffled’, and I used Amplify to get it to a reasonable level, but that made the applause between gags distort badly. Subsequent use of the Limiter helped significantly, but resulted in the introduction of some previously unnoticed background noise. All this was done randomly, without really understanding what I was doing, in terms of selecting Amplification, New Peak Amplitude (db), and all the various Limiter settings.

I am confused by the use of an optimum level of 0db, when the vertical baseline on the recorded waveform is 0db and varies between -1.0db and +1.0db. Can someone provide me with a simple explanation of the background to all of this, including about dbs and at what db level I can expect to get distortion, or direct me to a suitable tutorial, please?

Thanks.

0dB is the “digital maximum”*. If you go over 0dB you can get [u]clipping[/u] (distortion). On the (default) Audacity waveform display 0dB peaks will hit +1 and -1.

Audacity uses floating-point internally so it can go over 0dB without clipping.

Regular (integer) WAV files, digital-to-analog converters (playback), analog-to-digital converters (recording), and CDs are all hard-limited to 0dB.

This means that if your file goes over 0dB in Audacity but you are not playing at “full digital volume”, you may not be clipping your DAC (digital-to-analog converter) so you may not hear the distortion until you export the file.

It also means that if you use a format that can go over 0dB, you should still keep your peak down to 0dB or less because if the file goes over 0dB the listener’s DAC can clip.

The Amplify effect will default to whatever (positive or negative) gain is needed for “maximized” 0dB peaks. (This is also called “normalization”.)

BTW - It’s a good idea to Amplify or Normalize as the last step before exporting because effects/editing/processing can sometimes push the levels into clipping. (You can also run Amplify just to check the peaks, and then cancel the effect if you don’t actually want to make a volume change.)

Peak levels don’t correlate well with perceived loudness… If you normalize all of your recordings for 0dB peaks they won’t all be the same loudness.

Most commercial recordings are compressed and limited to bring-up the overall loudness without clipping. This can be tricky for amateurs because compression & limiting are a kind of distortion and they can sound like distortion or add unwanted side-effects.

but the maximum recording levels achievable did not exceed 12.

I assume you mean -12dB. That’s fine. It’s OK to boost digitally after recording. If you’ve done any analog recording you know that you want a “hot” signal to overcome tape noise, but with digital there is no tape noise.** Also, tape “soft clips” as you go over 0dB (and it can go over 0dB) so it’s OK to go occasionally-slightly “into the red”. Digital hard-clips at 0dB so it’s a good idea to leave plenty of headroom.

On playing back, the voice was low in volume and ‘muffled’,

Is the sound muffled when you just listen to the tape playback? …If the digitized audio is degraded compared to the original analog sound, we need to talk about your setup because that shouldn’t happen.

You can use EQ (Equalization) to bring-up the high frequencies and/or cut the low frequencies and hopefully make it more “clear”, but boosting the highs will also boost any tape hiss.

and I used Amplify to get it to a reasonable level, but that made the applause between gags distort badly.

If you don’t check the “Allow Clipping” box, Amplify won’t allow you to push the peaks into clipping.

Subsequent use of the Limiter helped significantly, but resulted in the introduction of some previously unnoticed background noise.

Yes… Limiting by itself doesn’t increase noise, but make-up gain (or otherwise boosting the levels) will bring-up the noise along with everything else. The same thing will happen if you crank-up the volume at playback time, so there’s no harm in doing it digitally.

You can also use the Envelope Tool to “manually” fade-down the volume during the applause. That’s more work but it will probably work better than limiting. (The trick with the Envelope Tool is to “fade” so you’re not making any sudden up or down jumps in volume.)

But again, if you fade-down the applause and then increase the overall volume, you’ll boost the noise.

You can try the Noise Reduction effect, but this really works best if you have a constant low-level background noise. …It works best when you don’t really need it. If the noise is bad (like the noise you likely have on a “live” recording), “The cure can be worse than the disease.”



\

  • 0dB is the highest you can “count” with a given number of bits. Everything is automatically scaled so a 24-bit file is not louder than an 8-bit file. With more bits, the increased dynamic range comes on the quiet-side. (With floating point audio 0dB has a numeric value of 1.0 and for all practical purposes there is no upper or lower limit.)

** There is something called digital quantization noise, but it’s more than 90dB down at 16-bits so we normally can’t hear it, and it’s usually way-quieter than the analog noise.

Thanks very much for your prompt and detailed reply - most helpful.

Your comment about perceived loudness is interesting. In order to get what I considered a reasonable volume to listen to, using Amplify with clipping allowed resulted in the distortion during applause. With no clipping allowed, the volume was not considered to be enough, and so my quest to reduce just the clipped applause in one go for the whole recording began.

Is the sound muffled when you just listen to the tape playback? Sadly I cannot comment on this, because the original recording was made at 1 7/8 ips whereas my current recorder (a vintage Grundig TK140, and valved at that) just uses 3 3/4 ips, and I therefore changed the speed within Audacity by 50% after the initial recording. It is quite likely that the original recording was far from optimal due to limitations imposed by the equipment used at the time (late 1960s), which I cannot 100% recall. Not only that but the DIN output from the recorder is volume dependant ie not a true line-out, and I had to use maximum to get a decent recording level! As I’m using Audacity on a laptop there is no direct line in, and I’m using a Behringer interface into the mic jack, so not an ideal setup at all! I should state that I am obviously aware that I cannot expect perfection here, but nevertheless wish if at all possible to improve the listening experience!
It may be that experimenting with the various filters available may improve the situation.

It sounds to me that using the envelope tool may be the best option regarding the distorted applause.

I’ll report back when I’ve had the opportunity to do so, but unfortunately relatively minor kitchen modifications have to take priority on my time!

I think that probably is.
Unfortunately there’s no good solution if the audience starts applauding loudly before the end of the music (hence the popularity of studio recordings without an audience).