Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Spectral Analysis of mp3 (Read 8310 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Spectral Analysis of mp3

So, I'm looking to buy a subwoofer to accompany my 2-ch speakers.  I know humans can hear down to 20Hz and feel lower than that.  In looking at subwoofer manufacturer frequency responses I decided it would be "fun" to look at some of my mp3 collection to see how low they go (no point in paying more for lower frequency response if, in truth, my known bass heavy music doesn't go down that far).

I have played around with some spectral analysis tools. Rather than looking at lines (which tend to move very quickly) it would be nice to see some raw data i.e. at 3 min 5 secs the 24 Hz freq was played at 120 db. Does anyone know of any software which will do this? I looked at waterfall spectrograms too be they seem to be even more difficult to read due to their use of colour to show intensity (pretty to look at though).

Secondly, from what I have seen it is common for all frequencies to be played at some level throughout a song (at least in my music). How do I know what db translates into something which would be heard / felt at a reasonable level when played by a sub?

I fugured this was the correct forum to post this in - sorry if I'm wrong about that.

Spectral Analysis of mp3

Reply #1 –
Quote from: AlphaWave on
I have played around with some spectral analysis tools. Rather than looking at lines (which tend to move very quickly) it would be nice to see some raw data i.e. at 3 min 5 secs the 24 Hz freq was played at 120 db. Does anyone know of any software which will do this? I looked at waterfall spectrograms too be they seem to be even more difficult to read due to their use of colour to show intensity (pretty to look at though).


Octave/matlab would be my choice.  This sort of thing is pretty easy, take FFTs with a sliding window, then sum the bin of interest across each FFT.  Then you can do whatever analysis (RMS, peak, percentile, etc) you want on the resulting data.

Quote from: AlphaWave on
Secondly, from what I have seen it is common for all frequencies to be played at some level throughout a song (at least in my music). How do I know what db translates into something which would be heard / felt at a reasonable level when played by a sub?


I don't know about that.  Usually music is said to have a spectrum somewhat more like pink noise (peaking at lower frequencies) than white noise.

But 10dB will have 10x the power and sound roughly 2x as loud (because hearing is not linear with intensity).

Spectral Analysis of mp3

Reply #2 –
Quote
(no point in paying more for lower frequency response if, in truth, my known bass heavy music doesn't go down that far).
Just FYI - The lowest note on a standard electric bass is about 40Hz.  A kick drum might have lower overtones.

If you look at pro subwoofers used for dance clubs & live music they are usually "tuned" to go down to around 40Hz.    The lower you go, the bigger your woofers need to be (or you need more of them) and the more amplifier power you need.    Around 40Hz seems to be a good compromise when you are trying to fill a large room with bass you can feel.

Of course, it takes a lot less power to fill a living room (or a car) with bass and you'll find home subs that to down to 20Hz or lower.

Quote
How do I know what db translates into something which would be heard / felt at a reasonable level when played by a sub?
If you want to check your setup in your room, you can generate some test tones, or pink noise, with Audacity. 

NOTE - Be careful with high-power test-tones....  A 100W speaker is designed for music with 100W peaks, and you might damage it with 100W test tones.      Be especially careful with high frequencies.  You might fry the tweeter in 100W speaker with a 20W continuous test-tone.

Quote
...the 24 Hz freq was played at 120 db.
120dB sounds like an SPL measurement (sound pressure level).  For that you need an SPL meter, or an acoustic spectrum analyzer because obviously the computer doesn't know how loud your speakers are, or if you speakers can even reproduce 24Hz.    You can find SPL meters for under $50 USD, but a cheapo meter may not be accurate from 20Hz-20kHz.   

Spectral Analysis of mp3

Reply #3 –
Quote from: AlphaWave on
...
I have played around with some spectral analysis tools. Rather than looking at lines (which tend to move very quickly) it would be nice to see some raw data i.e. at 3 min 5 secs the 24 Hz freq was played at 120 db. Does anyone know of any software which will do this? I looked at waterfall spectrograms too be they seem to be even more difficult to read due to their use of colour to show intensity (pretty to look at though).
...


I second saratoga's matlab/octave suggestion, but if you don't want to go that deep yet I suggest using Audacity or one of the other programs that plots levels versus time. To use Audacity, you can do the following:
- Obtain and install Audacity 2.0.4 or newer (Free download)
- Select Edit --> Preferences --> Spectrograms
- Select Window Size 32768
- Set: Minimum Frequency 0, Maximum Frequency 100, Gain 10, Range 40, Frequency Gain 0
  (You can tweak the values to make the resulting display better match the subjective sound of the bass.)
- Import a track (File --> Import, or Ctrl-Shift-I, select track)
The track will load and display the waveform.
- Click on the track title bar and select Spectrogram from the drop-down list.
Any significant content below 100 Hz should be clearly visible.
For additional analysis:
- Use the selection tool to select a few seconds of the audio.
- Select Analyse --> Plot Spectrum
- Change the Size to 65536
Read the graph from right to left. As a general rule, the levels should slope up from right to left as the frequency gets lower and lower. Eventually, the level will drop off again. Once it starts to drop off, you can ignore any lower frequencies - they won't be audible. The lower the frequency, the louder the signal has to be to be audible. 



Regards,
   Don Hills
"People hear what they see." - Doris Day

Spectral Analysis of mp3

Reply #4 –
Many thanks to all of you for your thoughts. Matlab/Octave do look a little advanced for my current ability so I will try Audacity first and do some learning if that turns out not to meet my needs.