Josef Rebbe

I'm working on a project where I need to tune recordings of singing bowls and low gongs to match exact frequencies (harmonics of certain spectrums).

Therefore I need a software that is able to analyse the recorded sounds in an exact way. It should output the exact frequencies and amplitude of all sine tones included in the recording.

Does anybody know a (preferably free) software that does reliable offline analysis of a soundfile (including low frequencies)?

It doesn't have to be in max necessarily.

I'm working on a project where I need to  tune recordings of singing bowls and low gongs to match exact frequencies (harmonics of certain spectrums).
Therefore I need a software that is able to analyse the recorded sounds in an exact way. It should output the exact frequencies and amplitude of all sine tones included in the recording.
Does anybody know a (preferably free) software that does reliable offline analysis of a soundfile (including low frequencies)?
It doesn't have to be in max necessarily.

exact-measurement-of-low-frequencies

the good thing about analysis of the lower frequencies in gongs is they last a long time, and don't change (much) after the initial attack. this means you can do a large - window fft analysis, say 8192 window size at 48khz, and get a frequency resolution of around 3 hz (ie error of +- 1.5 hz).

this may not seem "exact", but in audio analysis there is no such thing as "exact", only measurement with a particular degree of error.

you can do this in max-- I'm not sure what the largest fft window size is in max but you could experiment to find out.

if you want it more exact than that ( ie fractions of a hz ) you'd need to first approximately estimate via fft, then isolate a freq range withing the margin of error using a custom-made phase-linear filter, then dial up an 'analysis tone' (sine wave) to compare with the signal until the interference pattern (beating) goes away;

another way might be to down-sample the sounds, but analyse at the original sample rate to increase freq resolution in the lower frequencies for a given fft window size

the good thing about analysis of the lower frequencies in gongs is they last a long time, and don't change (much) after the initial attack. this means you can do a large - window fft analysis, say 8192 window size at 48khz, and get a frequency resolution of around 3 hz (ie error of +- 1.5 hz).
this may not seem "exact", but in audio analysis there is no such thing as "exact", only measurement with a particular degree of error.
you can do this in max-- I'm not sure what the largest fft window size is in max but you could experiment to find out.

if you want it more exact than that ( ie fractions of a hz ) you'd need to first approximately estimate via fft, then isolate a freq range withing the margin of error using a custom-made phase-linear filter, then dial up an 'analysis tone' (sine wave) to compare with the signal until the interference pattern (beating) goes away;
you can also do this in max

another way might be to down-sample the sounds, but analyse at the original sample rate to increase freq resolution in the lower frequencies for a given fft window size


You could use Audacity (menu Analyze -> Plot Spectrum). As pointed by Floating Point, you'll need a large window size to get information on the low range.

unfortunately Audacity doesn't work on mac OSX. Do you have any other suggestion?

I am no fan of audacity, but it works on OSX.

Software that does this task precisely is melodyne editor.

I am no fan of audacity, but it works on OSX.
Software that does this task precisely is melodyne editor.


exact measurement of low frequencies