I'm creating a waveform representation of a wav file and was curious what the best possible methods to go about this efficiently in python using the standard library. Some audio files could be minutes long.
Thanks!
5
-
1FWIW, numpy isn't part of the stdlib.mgilson– mgilson2013-05-21 21:29:17 +00:00Commented May 21, 2013 at 21:29
-
my bad. Didn't know that. I guess standard library with the except of numpy.chribis– chribis2013-05-21 21:34:34 +00:00Commented May 21, 2013 at 21:34
-
What exactly do you mean by "creating a waveform representation"? You mean a graphical representation of the waveform (i.e., a graph of sample value vs. time)? If so, are you looking to generate a 200000x200 PNG file, to draw an animated graph in a Tk window in real-time, or… ?abarnert– abarnert2013-05-21 21:36:25 +00:00Commented May 21, 2013 at 21:36
-
1.bp.blogspot.com/-GV_aYt3elV0/T84nErDNJ3I/AAAAAAAAAWI/… Something like this. This is for representation in autodesk maya. It will be a animation curve.chribis– chribis2013-05-21 21:39:10 +00:00Commented May 21, 2013 at 21:39
-
1Not stdlib, but I'd still consider looking at combining audiolab (docs) and matplotlib.Lukas Graf– Lukas Graf2013-05-21 21:50:16 +00:00Commented May 21, 2013 at 21:50
Add a comment
|
2 Answers
http://docs.python.org/2/library/wave.html - stdlib for reading wav files. A simple example of using it.
Off the top of my head this is how I'd do it (pseudocode)
fmts = (None, "=B", "=h", None, "=l")
fmt = fmts[sampwidth]
dcs = (None, 128, 0, None, 0)
dc = dcs[sampwidth]
image_width = 600
image_height = 300
chunk_size = len(wavefile.getnframes()) / image_width
def unpacker(frame):
return struct.unpack(fmt, frame)[0]
for i in range(chunk_size):
value = math.avg([unpacker(x) for x in wavefile.read_frames(chunk_size)])
# and then use value * 300 to figure out the vertical position for the pixel.
There are a variety of libraries you could use instead of writing an image, svg for example.
2 Comments
synthesizerpatel
Additionally: An example of a module that includes C code (that could Easily be converted to Python code instead..) github.com/superjoe30/PyWaveform
chribis
This is awesome, but i'm really just in need of an amplitude array from a wav file. I'm pretty sure I can work from here though!
The TimeSide library solves this problem in a very simple way, while providing some more interesting graphs.
I had this same problem and implemented the algorithm suggested by synthesizerpatel, later I discovered this library and got some better results.
1 Comment
lfagundes
Ah, missed the part you said "standard library". Anyway, this question is one of the first resources I found, hope TimeSide solves someone's problem.