Reply
 
Thread Tools Display Modes
  #1   Report Post  
Posted to rec.audio.pro
Ralph Barone[_3_] Ralph Barone[_3_] is offline
external usenet poster
 
Posts: 60
Default MP3 subharmonic synthesizer

OK, disregarding WHY you want to do such a thing, I'm trying to figure out
if this is possible.

As far as I can tell, MP3 encoding breaks the audio up into short chunks,
then does an FFT to convert into frequency/magnitude pairs, does some magic
to throw out theoretically inaudible frequencies, then writes the result to
a file in a compressed format. So, given that, shouldn't it be relatively
trivial to write software that takes in an MP3 file or stream, calculates
new frequency/magnitude pairs based on the incoming data, and then writes
the resulting data to a higher bit rate file/stream. If this is possible,
then things like subharmonic synthesizers and "aural exciters" (ie: adding
certain harmonics) should be pretty easy to implement inside the MP3
decoding process. Have I missed something?
  #2   Report Post  
Posted to rec.audio.pro
geoff geoff is offline
external usenet poster
 
Posts: 1,812
Default MP3 subharmonic synthesizer

On 29/10/2017 4:52 PM, Ralph Barone wrote:
OK, disregarding WHY you want to do such a thing, I'm trying to figure out
if this is possible.

As far as I can tell, MP3 encoding breaks the audio up into short chunks,
then does an FFT to convert into frequency/magnitude pairs, does some magic
to throw out theoretically inaudible frequencies, then writes the result to
a file in a compressed format. So, given that, shouldn't it be relatively
trivial to write software that takes in an MP3 file or stream, calculates
new frequency/magnitude pairs based on the incoming data, and then writes
the resulting data to a higher bit rate file/stream. If this is possible,
then things like subharmonic synthesizers and "aural exciters" (ie: adding
certain harmonics) should be pretty easy to implement inside the MP3
decoding process. Have I missed something?



Yes. Why would you want to do that - make crap sound even worse ?

geoff
  #3   Report Post  
Posted to rec.audio.pro
Scott Dorsey Scott Dorsey is offline
external usenet poster
 
Posts: 16,853
Default MP3 subharmonic synthesizer

geoff wrote:
On 29/10/2017 4:52 PM, Ralph Barone wrote:
As far as I can tell, MP3 encoding breaks the audio up into short chunks,
then does an FFT to convert into frequency/magnitude pairs, does some magic
to throw out theoretically inaudible frequencies, then writes the result to
a file in a compressed format. So, given that, shouldn't it be relatively
trivial to write software that takes in an MP3 file or stream, calculates
new frequency/magnitude pairs based on the incoming data, and then writes
the resulting data to a higher bit rate file/stream. If this is possible,
then things like subharmonic synthesizers and "aural exciters" (ie: adding
certain harmonics) should be pretty easy to implement inside the MP3
decoding process. Have I missed something?


This is kind of an oversimplified view of the process, but in fact there is
a frequency-domain representation of part of the signal in the mp3 file
and consequently that sort of trickery is easy to implement on the encoded
file.

The original MP3 patent is worth reading, it's pretty dense and obfuscated but
it is detailed.

Yes. Why would you want to do that - make crap sound even worse ?


A lot of people do. And some embedded MP3 players do in fact. It is a
sad world that we live in.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."
  #4   Report Post  
Posted to rec.audio.pro
Les Cargill[_4_] Les Cargill[_4_] is offline
external usenet poster
 
Posts: 1,383
Default MP3 subharmonic synthesizer

Ralph Barone wrote:
OK, disregarding WHY you want to do such a thing, I'm trying to
figure out if this is possible.

As far as I can tell, MP3 encoding breaks the audio up into short
chunks, then does an FFT to convert into frequency/magnitude pairs,
does some magic to throw out theoretically inaudible frequencies,
then writes the result to a file in a compressed format. So, given
that, shouldn't it be relatively trivial to write software that takes
in an MP3 file or stream, calculates new frequency/magnitude pairs
based on the incoming data, and then writes the resulting data to a
higher bit rate file/stream. If this is possible, then things like
subharmonic synthesizers and "aural exciters" (ie: adding certain
harmonics) should be pretty easy to implement inside the MP3 decoding
process. Have I missed something?


The whole thing of tying this to the MP3 encoding process is a total
McGuffin/red herring.

I have wasted more than a few hours with FFTW and other libraries
playing with stuff like this. Effectively, you can reproduce the .MP3
"experience" by performing an FFT on a .wav file, then discarding the
N buckets with the least energy. It's not the exact same thing, but it's
like that.

That's a completely different thing from pitch shifting or subharmonic
generation. But the short answer is "no, you cant just naively play
with the FFT buckets and get useable pitch changes." There exist
transforms for stuff like this but they're nontrivial - on the order of
things like a phase vocoder.

So you'd think you could just assign the sum of buckets N and N+2 to
bucket N/2 ( for the appropriate values of N ) and get an octave down,
right?

Yes, but it also halves the playback rate - it's exactly like playing
back a tape at half speed. And if you'll think about it, that actually
makes perfect sense.

--
Les Cargill
  #5   Report Post  
Posted to rec.audio.pro
Les Cargill[_4_] Les Cargill[_4_] is offline
external usenet poster
 
Posts: 1,383
Default MP3 subharmonic synthesizer

geoff wrote:
On 29/10/2017 4:52 PM, Ralph Barone wrote:
OK, disregarding WHY you want to do such a thing, I'm trying to
figure out if this is possible.

As far as I can tell, MP3 encoding breaks the audio up into short
chunks, then does an FFT to convert into frequency/magnitude pairs,
does some magic to throw out theoretically inaudible frequencies,
then writes the result to a file in a compressed format. So, given
that, shouldn't it be relatively trivial to write software that
takes in an MP3 file or stream, calculates new frequency/magnitude
pairs based on the incoming data, and then writes the resulting
data to a higher bit rate file/stream. If this is possible, then
things like subharmonic synthesizers and "aural exciters" (ie:
adding certain harmonics) should be pretty easy to implement inside
the MP3 decoding process. Have I missed something?



Yes. Why would you want to do that - make crap sound even worse ?

geoff


Worse is better.

--
Les Cargill


  #6   Report Post  
Posted to rec.audio.pro
Ralph Barone[_3_] Ralph Barone[_3_] is offline
external usenet poster
 
Posts: 60
Default MP3 subharmonic synthesizer

Scott Dorsey wrote:
geoff wrote:
On 29/10/2017 4:52 PM, Ralph Barone wrote:
As far as I can tell, MP3 encoding breaks the audio up into short chunks,
then does an FFT to convert into frequency/magnitude pairs, does some magic
to throw out theoretically inaudible frequencies, then writes the result to
a file in a compressed format. So, given that, shouldn't it be relatively
trivial to write software that takes in an MP3 file or stream, calculates
new frequency/magnitude pairs based on the incoming data, and then writes
the resulting data to a higher bit rate file/stream. If this is possible,
then things like subharmonic synthesizers and "aural exciters" (ie: adding
certain harmonics) should be pretty easy to implement inside the MP3
decoding process. Have I missed something?


This is kind of an oversimplified view of the process, but in fact there is
a frequency-domain representation of part of the signal in the mp3 file
and consequently that sort of trickery is easy to implement on the encoded
file.

The original MP3 patent is worth reading, it's pretty dense and obfuscated but
it is detailed.


Does reading the Wikipedia page count :-)

Yes. Why would you want to do that - make crap sound even worse ?


A lot of people do. And some embedded MP3 players do in fact. It is a
sad world that we live in.
--scott


Humph. Every time I think I've come up with an original idea (no matter how
bad), it seems that somebody else always beats me to implementation.


  #7   Report Post  
Posted to rec.audio.pro
Ralph Barone[_3_] Ralph Barone[_3_] is offline
external usenet poster
 
Posts: 60
Default MP3 subharmonic synthesizer

Les Cargill wrote:
Ralph Barone wrote:
OK, disregarding WHY you want to do such a thing, I'm trying to
figure out if this is possible.

As far as I can tell, MP3 encoding breaks the audio up into short
chunks, then does an FFT to convert into frequency/magnitude pairs,
does some magic to throw out theoretically inaudible frequencies,
then writes the result to a file in a compressed format. So, given
that, shouldn't it be relatively trivial to write software that takes
in an MP3 file or stream, calculates new frequency/magnitude pairs
based on the incoming data, and then writes the resulting data to a
higher bit rate file/stream. If this is possible, then things like
subharmonic synthesizers and "aural exciters" (ie: adding certain
harmonics) should be pretty easy to implement inside the MP3 decoding
process. Have I missed something?


The whole thing of tying this to the MP3 encoding process is a total
McGuffin/red herring.


Actually, it was the decoding process that I was looking at, since the
encoded file already contains the data in a convenient frequency domain
representation (not that it matters any more in a world where computational
power is so cheap).

I have wasted more than a few hours with FFTW and other libraries
playing with stuff like this. Effectively, you can reproduce the .MP3
"experience" by performing an FFT on a .wav file, then discarding the
N buckets with the least energy. It's not the exact same thing, but it's
like that.

That's a completely different thing from pitch shifting or subharmonic
generation. But the short answer is "no, you cant just naively play
with the FFT buckets and get useable pitch changes." There exist
transforms for stuff like this but they're nontrivial - on the order of
things like a phase vocoder.

So you'd think you could just assign the sum of buckets N and N+2 to
bucket N/2 ( for the appropriate values of N ) and get an octave down,
right?


Pretty much. Do a weighted sum of the magnitudes of buckets N-1, N and N+1,
multiply that by a scaling factor and add that into bucket N/2.

Yes, but it also halves the playback rate - it's exactly like playing
back a tape at half speed. And if you'll think about it, that actually
makes perfect sense.


I've thought about it, and so far it doesn't make perfect sense. Since you
haven't actually changed the number of frames in the MP3 file (only the
contents), I'm not sure why the duration of playback would change.



  #8   Report Post  
Posted to rec.audio.pro
Les Cargill[_4_] Les Cargill[_4_] is offline
external usenet poster
 
Posts: 1,383
Default MP3 subharmonic synthesizer

Ralph Barone wrote:
Les Cargill wrote:
Ralph Barone wrote:

snip

I've thought about it, and so far it doesn't make perfect sense.
Since you haven't actually changed the number of frames in the MP3
file (only the contents), I'm not sure why the duration of playback
would change.




The actual frequencies change but the relationships between them don't
unless you do something more sophisticated. It works like a Doppler
shift - I'd guess ( without being able to prove it ) that it works
exactly like a Doppler shift.

I was mainly speeding things up, and in that case, you get the same
signal at twice the "speed" - and you get two copies of it That's
how it preserves "number of frames".


--
Les Cargill
Reply
Thread Tools
Display Modes

Posting Rules

Smilies are On
[IMG] code is On
HTML code is Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
DBX Subharmonic Synthesizer. PanHandler[_2_] Pro Audio 33 December 19th 08 08:19 AM
DBX Subharmonic Synthesizer PanHandler[_2_] Tech 0 December 13th 08 08:50 PM
FS: dbx 120 subharmonic synthesizer John Noll Pro Audio 7 September 3rd 03 05:05 PM
FS: dbx 120 subharmonic synthesizer John Noll Marketplace 8 September 3rd 03 05:05 PM


All times are GMT +1. The time now is 08:47 AM.

Powered by: vBulletin
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
Copyright ©2004-2024 AudioBanter.com.
The comments are property of their posters.
 

About Us

"It's about Audio and hi-fi"