24 Oct
2005
24 Oct
'05
4:22 p.m.
Hi,
I'm being really lazy here and looking for some ideas on how to do
a couple of things easily. I really don't want to spend much time on
there. Thanks in advance.
1) Assume I have an existing wave file (mono or stereo) and want to
convert it to text. I'd like each sample written in 8/16/24-bit hex to
its own line in the file. Any easy way to do it?
2) Assume I've chosen a sample rate and bit-depth. Now want to
generate a hex file with sine wave data, again with each sample
written to its own line. Any easy way to do it?
Thanks,
Mark
24 Oct
24 Oct
5:09 p.m.
New subject: [linux-audio-user] Digital audio files for hardware verification
On Mon, 2005-10-24 at 11:22 -0700, Mark Knecht wrote:
Hi,
I'm being really lazy here and looking for some ideas on how to do
a couple of things easily. I really don't want to spend much time on
there. Thanks in advance.
1) Assume I have an existing wave file (mono or stereo) and want to
convert it to text. I'd like each sample written in 8/16/24-bit hex to
its own line in the file. Any easy way to do it?
sox will do this, if you use the .dat output file type.
sox something.wav something.dat
not sure how to extract just 1 channel from a stereo file, i would
probably run the output through perl or something afterwards.
can't think of an easy way to do #2.
25 Oct
25 Oct
5:49 a.m.
On Mon, 24 Oct 2005 11:22:36 -0700 Mark Knecht <markknecht(a)gmail.com>
wrote:
2) Assume I've chosen a sample rate and bit-depth.
Now want to
generate a hex file with sine wave data, again with each sample
written to its own line. Any easy way to do it?
awk is your friend (at least, mine ;-), put the following command
in a single line:
gawk -vt=1 -vf=2 -vr=220 -vd=6
'BEGIN{z=lshift(1,d-1);w=(d-1)/4+1;for(i=0;i<=r*t;i++)printf("0x%0*x\n",w,z+0.5+(z-1)*sin((f/r)*i*2*3.1415926));exit}'
> outfile.dat
Arguments (-vVARIABLE=...)
VARIABLEs: t ... total time to generate/seconds
f ... frequency/Hz
r ... sample rate/Hz
d ... depth/bit
Output has an offset of 1<<(depth - 1).
--
Markus Schwarzenberg
11:54 a.m.
On 10/25/05, Markus Schwarzenberg <schwarzb(a)ipms.fraunhofer.de> wrote:
On Mon, 24 Oct 2005 11:22:36 -0700 Mark Knecht
<markknecht(a)gmail.com>
wrote:
Very cool Markus. Thanks. This is a great example of what Unix can do
at the command line.
Thanks very much,
Mark
2) Assume I've chosen a sample rate and
bit-depth. Now want to
generate a hex file with sine wave data, again with each sample
written to its own line. Any easy way to do it?
awk is your friend (at least, mine ;-), put the following command
in a single line:
gawk -vt=1 -vf=2 -vr=220 -vd=6
'BEGIN{z=lshift(1,d-1);w=(d-1)/4+1;for(i=0;i<=r*t;i++)printf("0x%0*x\n",w,z+0.5+(z-1)*sin((f/r)*i*2*3.1415926));exit}'
> outfile.dat
Arguments (-vVARIABLE=...)
VARIABLEs: t ... total time to generate/seconds
f ... frequency/Hz
r ... sample rate/Hz
d ... depth/bit
Output has an offset of 1<<(depth - 1).
--
Markus Schwarzenberg 
2:35 p.m.
Awk is powerful and concise, but it's not my friend - I can never remember how
the heck it works ;-)
I'd probably do like Paul D. suggests and use existing tools - e.g.
find something
to generate an appropriate soundfile, then feed it through sox.
But just for fun, here's something silly I just whipped up:
$ ./sine_hex.py
Usage: ./sine_hex.py frequency [sampling rate] [bits]
Print one cycle of a sine wave of approximately the
given frequency.
Values given in hex with *bits* precision at *sampling rate*.
Default sampling rate is 44100. Default bits is 16.
pwinkler@Winkler-P-LT2K ~
$ ./sine_hex.py 24000 48000
0000
ffff
pwinkler@Winkler-P-LT2K ~
$ ./sine_hex.py 3000 48000
0000
31f1
61f7
8e39
b504
d4da
ec82
fb14
ffff
fb14
ec82
d4da
b504
8e39
61f7
31f1
And here's the source. Most of it is argument handling :-)
http://www.slinkp.com/~paul/sine_hex.py
--
http://www.slinkp.com
3 p.m.
On 10/25/05, Paul Winkler <pw_lists(a)slinkp.com> wrote:
Awk is powerful and concise, but it's not my
friend - I can never remember how
the heck it works ;-)
I'd probably do like Paul D. suggests and use existing tools - e.g.
find something
to generate an appropriate soundfile, then feed it through sox.
But just for fun, here's something silly I just whipped up:
$ ./sine_hex.py
Usage: ./sine_hex.py frequency [sampling rate] [bits]
Print one cycle of a sine wave of approximately the
given frequency.
Values given in hex with *bits* precision at *sampling rate*.
Default sampling rate is 44100. Default bits is 16.
pwinkler@Winkler-P-LT2K ~
$ ./sine_hex.py 24000 48000
0000
ffff
pwinkler@Winkler-P-LT2K ~
$ ./sine_hex.py 3000 48000
0000
31f1
61f7
8e39
b504
d4da
ec82
fb14
ffff
fb14
ec82
d4da
b504
8e39
61f7
31f1
And here's the source. Most of it is argument handling :-)
http://www.slinkp.com/~paul/sine_hex.py
Intereing. Another country heard from.
One question that comes to mind, as I play with these solutions, is
deciding how to evaluate the outputs. For instance I was fiddling with
generating a sawtooth from Marcus' s method. I could do the same with
this, but how do I know each of them is really implementing what I'm
interested in?
What tools do people use for FFT analysis on files like this?
Cheers,
Mark
3:57 p.m.
This is a huge can of worms you're opening here ...
A simple sawtooth wave is going to have a lot of aliasing, unless you
specifically use an algorithm that produces a band-limited approximation,
at which point you've already left me far behind ;-)
For FFT, my needs have been very modest, I just eyeball stuff in Snd.
There is also www.baudline.com , very slick but it is merely free and not Free.
-PW
On 10/25/05, Mark Knecht <markknecht(a)gmail.com> wrote:
On 10/25/05, Paul Winkler <pw_lists(a)slinkp.com>
wrote:
--
http://www.slinkp.com
Awk is powerful and concise, but it's not my
friend - I can never remember how
the heck it works ;-)
I'd probably do like Paul D. suggests and use existing tools - e.g.
find something
to generate an appropriate soundfile, then feed it through sox.
But just for fun, here's something silly I just whipped up:
$ ./sine_hex.py
Usage: ./sine_hex.py frequency [sampling rate] [bits]
Print one cycle of a sine wave of approximately the
given frequency.
Values given in hex with *bits* precision at *sampling rate*.
Default sampling rate is 44100. Default bits is 16.
pwinkler@Winkler-P-LT2K ~
$ ./sine_hex.py 24000 48000
0000
ffff
pwinkler@Winkler-P-LT2K ~
$ ./sine_hex.py 3000 48000
0000
31f1
61f7
8e39
b504
d4da
ec82
fb14
ffff
fb14
ec82
d4da
b504
8e39
61f7
31f1
And here's the source. Most of it is argument handling :-)
http://www.slinkp.com/~paul/sine_hex.py
Intereing. Another country heard from.
One question that comes to mind, as I play with these solutions, is
deciding how to evaluate the outputs. For instance I was fiddling with
generating a sawtooth from Marcus' s method. I could do the same with
this, but how do I know each of them is really implementing what I'm
interested in?
What tools do people use for FFT analysis on files like this?
Cheers,
Mark
7027
days inactive
7028
days old
linux-audio-user@lists.linuxaudio.org
6 comments
4 participants
participants (4)
-
Mark Knecht -
Markus Schwarzenberg -
Paul Davis -
Paul Winkler