noise measurements

Hello all!

I have been using Audacity for general editing work of music and speeches for quite a sometime.
Now I have a new task at the workplace. It is a welding workshop. We have to re-arrange the welding and grinding areas and have to keep noise level below 85 dBs. I have recorded few samples on different dates as the work progressed. Can I use Audacity to compare those samples to find out what outfit gave the minimum noise level.

Thanking you in anticipation…


To get an absolute measurement you need to use a calibrated system.
The recorded level in Audacity depends on the type, position and sensitivity of the microphone, and the recording level set in Audacity, and the recording level set in the operating system, and the characteristics of the sound card, and many other things.

What you can do with Audacity is that you can set up your computer and microphone in one place, record something, then move the equipment around and record again, WITHOUT moving the microphone or changing ANY settings on the computer or Audacity, and then record again. It is essential that nothing changes in the recording set-up. Then you can play back the recordings and see if one recording sounds louder than the other. This is totally unscientific, but it may give you a general idea,

To get proper measurements you need to use a dB meter and use it in accordance with the instructions. Noise measurement is a complex business.

[u]Sound Level Meter[/u] Check your regulations before choosing one. You might need an A-Weighed SPL meter like this one.

If you ever end-up in a “legal situation”, you’d need to have your SPL meter calibrated periodically by an independent lab (which will cost at least as much as a cheap SPL meter), and you’d need to take and log measurements regularly under particular conditions. If you just want to be generally safe and in-compliance, that shouldn’t be necessary.

Thank you all for the prompt replies.
Please clarify following too.

  1. In an Audacity help manual I saw a statement, “Smaller the negative number louder the noise”.
  2. In the Audacity plot spectrum you get -ve numbers. How they are compared with “85dB”?
  3. I saw the facility to obtain a txt file from the spectrum. If I combine all txt data from my recordings at different workshop arrangments to a single spreadsheet and plot, would they direct me to the best outfit?

All this is done in preparation before the legal issue.

Once again, eagerly waiting for a reply…


Those numbers won’t help you much.
Again, The values are only relative ratios to the maximum recording level that Audacity can handle.
This means that your measurement must be compared to a signal that has 85 dB.
You have to set up your equipment near the noise source you want to compare.
You’ll then record a 85 dB calibrated noise sound - radiating from exact the same point as your target noise source.
In Audacity this signal may have a highest peak of -10 dB after recording.
If you then record the real noise source - same settings as before - the peak could be -20 dB and thus you know that you’re below the limit.
However, this isn’t absolute and you can’t say “my noise is 10 dB quieter than the 85 dB limit”.
Although there at least to plug-ins in the Nyquist archive which make use of the a-weighted filter curve, it isn’t recommendable to use them as a replacement for a calibrated sound level meter.
The question is after all, where should the 85 dB calibrated noise come from?

“dB” is a “ratio” that compares one signal level to a reference level.

Typically, when talking about “sound pressure”, the reference level is defined as the level at which the sound source is at the threshold of being audible, given a specified frequency range and a specified distance from the sound source and in echo free conditions. All audible sounds will thus have positive values (greater than 0 dB).

For “signal levels” in equipment (including recording applications such as Audacity), 0 dB is defined as the maximum peak signal level (full scale). Thus all valid signals are negative (less than 0 dB).

There is no direct relationship between the signal level in Audacity and the sound level of the thing that is being recorded.
A louder sound will produce a bigger signal in Audacity, provided that the same microphone is used, in the same position relative to the sound source, in the same environment and with the same input level settings in Audacity, in the operating system, in the microphone pre-amp, and in any other equipment in the signal chain.

I think you need to get a proper “sound level meter”.

Hi all,

Thanks for educating me on these numbers.

Now I understand that a “meter” is required. Auditors definitely will bring one! There are Android apps which claim to do the trick. May be I can

But, to plan the grinding booths, select suitable absorbing material some sort of a measure is required. I think Audacity has provided me with that. As I was suggesting about the spread sheet I did a rough one, delighted with what I got, which had data collected from 27 Feb till yesterday. Today I repeated all under similar circumstances using the laptop mike but one hand grinder.

We were provided with a concrete roof over the grinding area. Every body noticed that it increased the ‘noise’. Then it was shifted to another section under a plywood roof. Immediately I noticed a remarkable reduction in noise. Thereafter we ordered more plywood sheets, made temporary enclosure and took measurements and plotted the attached file. The data are first two columns under concrete roof, next three under plywood roof, inside and outside the temp booth.

May be the attachment do not have any legal significance, but can convince the management in which direction we must focus to cut down costs also.

Please comment about the content of this posting and any recommendations for future.