"""Routines to help recognizing sound files.
Function whathdr() recognizes various types of sound file headers.
It understands almost all headers that SOX can decode.
The return tuple contains the following items, in this order:
- file type (as SOX understands it)
- sampling rate (0 if unknown or hard to decode)
- number of channels (0 if unknown or hard to decode)
- number of frames in the file (-1 if unknown or hard to decode)
- number of bits/sample, or 'U' for U-LAW, or 'A' for A-LAW
If the file doesn't have a recognizable type, it returns None.
If the file can't be opened, IOError is raised.
To compute the total time, divide the number of frames by the
sampling rate (a frame contains a sample for each channel).
Function what() calls whathdr(). (It used to also use some
heuristics for raw data, but this doesn't work very well.)
Finally, the function test() is a simple main program that calls
what() for all files mentioned on the argument list. For directory
arguments it calls what() for all files in that directory. Default
argument is "." (testing all files in the current directory). The
option -r tells it to recurse down directories found inside
explicitly given directories.
# The file structure is top-down except that the test program and its
__all__ = ["what","whathdr"]
"""Guess the type of a sound file"""
"""Recognize sound headers"""
#-----------------------------------#
# Subroutines per sound header type #
#-----------------------------------#
except (EOFError, aifc.Error):
return (fmt, a.getframerate(), a.getnchannels(), \
a.getnframes(), 8*a.getsampwidth())
elif h[:4] in ('\0ds.', 'dns.'):
sample_size = 1 # default
frame_size = sample_size * nchannels
return type, rate, nchannels, data_size//frame_size, sample_bits
if h[65:69] != 'FSSD' or h[128:132] != 'HCOM':
divisor = get_long_be(h[128+16:128+20])
return 'hcom', 22050//divisor, 1, -1, 8
if h[:20] != 'Creative Voice File\032':
sbseek = get_short_le(h[20:22])
if 0 <= sbseek < 500 and h[sbseek] == '\1':
ratecode = ord(h[sbseek+4])
rate = int(1000000.0 / (256 - ratecode))
return 'voc', rate, 1, -1, 8
# 'RIFF' <len> 'WAVE' 'fmt ' <len>
if h[:4] != 'RIFF' or h[8:12] != 'WAVE' or h[12:16] != 'fmt ':
style = get_short_le(h[20:22])
nchannels = get_short_le(h[22:24])
rate = get_long_le(h[24:28])
sample_bits = get_short_le(h[34:36])
return 'wav', rate, nchannels, -1, sample_bits
if h[:4] != 'FORM' or h[8:12] != '8SVX':
# Should decode it to get #channels -- assume always 1
return '8svx', 0, 1, 0, 8
nsamples = get_long_le(h[8:12])
rate = get_short_le(h[20:22])
return 'sndt', rate, 1, nsamples, 8
rate = get_short_le(h[2:4])
if 4000 <= rate <= 25000:
return 'sndr', rate, 1, -1, 8
#---------------------------------------------#
# Subroutines to extract numbers from strings #
#---------------------------------------------#
return (ord(s[0])<<24) | (ord(s[1])<<16) | (ord(s[2])<<8) | ord(s[3])
return (ord(s[3])<<24) | (ord(s[2])<<16) | (ord(s[1])<<8) | ord(s[0])
return (ord(s[0])<<8) | ord(s[1])
return (ord(s[1])<<8) | ord(s[0])
if sys.argv[1:] and sys.argv[1] == '-r':
testall(sys.argv[1:], recursive, 1)
testall(['.'], recursive, 1)
except KeyboardInterrupt:
sys.stderr.write('\n[Interrupted]\n')
def testall(list, recursive, toplevel):
if os.path.isdir(filename):
if recursive or toplevel:
names = glob.glob(os.path.join(filename, '*'))
testall(names, recursive, 0)
print '*** directory (use -r) ***'
print '*** not found ***'
if __name__ == '__main__':