403 lines
14 KiB
Python
403 lines
14 KiB
Python
|
# Copyright (c) 2013 The Chromium Authors. All rights reserved.
|
||
|
|
# Use of this source code is governed by a BSD-style license that can be
|
||
|
|
# found in the LICENSE file.
|
||
|
|
|
||
|
|
import logging
|
||
|
|
import re
|
||
|
|
import subprocess
|
||
|
|
|
||
|
|
from autotest_lib.client.cros.audio import cmd_utils
|
||
|
|
|
||
|
|
SOX_PATH = 'sox'
|
||
|
|
|
||
|
|
def _raw_format_args(channels, bits, rate):
|
||
|
|
"""Gets raw format args used in sox.
|
||
|
|
|
||
|
|
@param channels: Number of channels.
|
||
|
|
@param bits: Bit length for a sample.
|
||
|
|
@param rate: Sampling rate.
|
||
|
|
|
||
|
|
@returns: A list of args.
|
||
|
|
|
||
|
|
"""
|
||
|
|
args = ['-t', 'raw', '-e', 'signed']
|
||
|
|
args += _format_args(channels, bits, rate)
|
||
|
|
return args
|
||
|
|
|
||
|
|
|
||
|
|
def _format_args(channels, bits, rate):
|
||
|
|
"""Gets format args used in sox.
|
||
|
|
|
||
|
|
@param channels: Number of channels.
|
||
|
|
@param bits: Bit length for a sample.
|
||
|
|
@param rate: Sampling rate.
|
||
|
|
|
||
|
|
@returns: A list of args.
|
||
|
|
|
||
|
|
"""
|
||
|
|
return ['-c', str(channels), '-b', str(bits), '-r', str(rate)]
|
||
|
|
|
||
|
|
|
||
|
|
def generate_sine_tone_cmd(
|
||
|
|
filename, channels=2, bits=16, rate=48000, duration=None, frequencies=440,
|
||
|
|
gain=None, vol=None, raw=True):
|
||
|
|
"""Gets a command to generate sine tones at specified ferquencies.
|
||
|
|
|
||
|
|
@param filename: The name of the file to store the sine wave in.
|
||
|
|
@param channels: The number of channels.
|
||
|
|
@param bits: The number of bits of each sample.
|
||
|
|
@param rate: The sampling rate.
|
||
|
|
@param duration: The length of the generated sine tone (in seconds).
|
||
|
|
@param frequencies: The frequencies of the sine wave. Pass a number or a
|
||
|
|
list to specify frequency for each channel.
|
||
|
|
@param gain: The gain (in db).
|
||
|
|
@param vol: A float for volume scale used in sox command.
|
||
|
|
E.g. 1.0 is the same. 0.5 to scale volume by
|
||
|
|
half. -1.0 to invert the data.
|
||
|
|
@param raw: True to use raw data format. False to use what filename specifies.
|
||
|
|
|
||
|
|
"""
|
||
|
|
args = [SOX_PATH, '-n']
|
||
|
|
if raw:
|
||
|
|
args += _raw_format_args(channels, bits, rate)
|
||
|
|
else:
|
||
|
|
args += _format_args(channels, bits, rate)
|
||
|
|
args.append(filename)
|
||
|
|
args.append('synth')
|
||
|
|
if duration is not None:
|
||
|
|
args.append(str(duration))
|
||
|
|
if not isinstance(frequencies, list):
|
||
|
|
frequencies = [frequencies]
|
||
|
|
for freq in frequencies:
|
||
|
|
args += ['sine', str(freq)]
|
||
|
|
if gain is not None:
|
||
|
|
args += ['gain', str(gain)]
|
||
|
|
if vol is not None:
|
||
|
|
args += ['vol', str(vol)]
|
||
|
|
return args
|
||
|
|
|
||
|
|
|
||
|
|
def noise_profile(*args, **kwargs):
|
||
|
|
"""A helper function to execute the noise_profile_cmd."""
|
||
|
|
return cmd_utils.execute(noise_profile_cmd(*args, **kwargs))
|
||
|
|
|
||
|
|
|
||
|
|
def noise_profile_cmd(input, output, channels=1, bits=16, rate=48000):
|
||
|
|
"""Gets the noise profile of the input audio.
|
||
|
|
|
||
|
|
@param input: The input audio.
|
||
|
|
@param output: The file where the output profile will be stored in.
|
||
|
|
@param channels: The number of channels.
|
||
|
|
@param bits: The number of bits of each sample.
|
||
|
|
@param rate: The sampling rate.
|
||
|
|
"""
|
||
|
|
args = [SOX_PATH]
|
||
|
|
args += _raw_format_args(channels, bits, rate)
|
||
|
|
args += [input, '-n', 'noiseprof', output]
|
||
|
|
return args
|
||
|
|
|
||
|
|
|
||
|
|
def noise_reduce(*args, **kwargs):
|
||
|
|
"""A helper function to execute the noise_reduce_cmd."""
|
||
|
|
return cmd_utils.execute(noise_reduce_cmd(*args, **kwargs))
|
||
|
|
|
||
|
|
|
||
|
|
def noise_reduce_cmd(
|
||
|
|
input, output, noise_profile, channels=1, bits=16, rate=48000):
|
||
|
|
"""Reduce noise in the input audio by the given noise profile.
|
||
|
|
|
||
|
|
@param input: The input audio file.
|
||
|
|
@param output: The output file in which the noise reduced audio is stored.
|
||
|
|
@param noise_profile: The noise profile.
|
||
|
|
@param channels: The number of channels.
|
||
|
|
@param bits: The number of bits of each sample.
|
||
|
|
@param rate: The sampling rate.
|
||
|
|
"""
|
||
|
|
args = [SOX_PATH]
|
||
|
|
format_args = _raw_format_args(channels, bits, rate)
|
||
|
|
args += format_args
|
||
|
|
args.append(input)
|
||
|
|
# Uses the same format for output.
|
||
|
|
args += format_args
|
||
|
|
args.append(output)
|
||
|
|
args.append('noisered')
|
||
|
|
args.append(noise_profile)
|
||
|
|
return args
|
||
|
|
|
||
|
|
|
||
|
|
def extract_channel_cmd(
|
||
|
|
input, output, channel_index, channels=2, bits=16, rate=48000):
|
||
|
|
"""Extract the specified channel data from the given input audio file.
|
||
|
|
|
||
|
|
@param input: The input audio file.
|
||
|
|
@param output: The output file to which the extracted channel is stored
|
||
|
|
@param channel_index: The index of the channel to be extracted.
|
||
|
|
Note: 1 for the first channel.
|
||
|
|
@param channels: The number of channels.
|
||
|
|
@param bits: The number of bits of each sample.
|
||
|
|
@param rate: The sampling rate.
|
||
|
|
"""
|
||
|
|
args = [SOX_PATH]
|
||
|
|
args += _raw_format_args(channels, bits, rate)
|
||
|
|
args.append(input)
|
||
|
|
args += ['-t', 'raw', output]
|
||
|
|
args += ['remix', str(channel_index)]
|
||
|
|
return args
|
||
|
|
|
||
|
|
|
||
|
|
def stat_cmd(input, channels=1, bits=16, rate=44100):
|
||
|
|
"""Get statistical information about the input audio data.
|
||
|
|
|
||
|
|
The statistics will be output to standard error.
|
||
|
|
|
||
|
|
@param input: The input audio file.
|
||
|
|
@param channels: The number of channels.
|
||
|
|
@param bits: The number of bits of each sample.
|
||
|
|
@param rate: The sampling rate.
|
||
|
|
"""
|
||
|
|
args = [SOX_PATH]
|
||
|
|
args += _raw_format_args(channels, bits, rate)
|
||
|
|
args += [input, '-n', 'stat']
|
||
|
|
return args
|
||
|
|
|
||
|
|
|
||
|
|
def get_stat(*args, **kargs):
|
||
|
|
"""A helper function to execute the stat_cmd.
|
||
|
|
|
||
|
|
It returns the statistical information (in text) read from the standard
|
||
|
|
error.
|
||
|
|
"""
|
||
|
|
p = cmd_utils.popen(stat_cmd(*args, **kargs), stderr=subprocess.PIPE)
|
||
|
|
|
||
|
|
#The output is read from the stderr instead of stdout
|
||
|
|
stat_output = p.stderr.read()
|
||
|
|
cmd_utils.wait_and_check_returncode(p)
|
||
|
|
return parse_stat_output(stat_output)
|
||
|
|
|
||
|
|
|
||
|
|
_SOX_STAT_ATTR_MAP = {
|
||
|
|
'Samples read': ('sameple_count', int),
|
||
|
|
'Length (seconds)': ('length', float),
|
||
|
|
'RMS amplitude': ('rms', float),
|
||
|
|
'Rough frequency': ('rough_frequency', float)}
|
||
|
|
|
||
|
|
_RE_STAT_LINE = re.compile('(.*):(.*)')
|
||
|
|
|
||
|
|
class _SOX_STAT:
|
||
|
|
def __str__(self):
|
||
|
|
return str(vars(self))
|
||
|
|
|
||
|
|
|
||
|
|
def _remove_redundant_spaces(value):
|
||
|
|
return ' '.join(value.split()).strip()
|
||
|
|
|
||
|
|
|
||
|
|
def parse_stat_output(stat_output):
|
||
|
|
"""A helper function to parses the stat_cmd's output to get a python object
|
||
|
|
for easy access to the statistics.
|
||
|
|
|
||
|
|
It returns a python object with the following attributes:
|
||
|
|
.sample_count: The number of the audio samples.
|
||
|
|
.length: The length of the audio (in seconds).
|
||
|
|
.rms: The RMS value of the audio.
|
||
|
|
.rough_frequency: The rough frequency of the audio (in Hz).
|
||
|
|
|
||
|
|
@param stat_output: The statistics ouput to be parsed.
|
||
|
|
"""
|
||
|
|
stat = _SOX_STAT()
|
||
|
|
|
||
|
|
for line in stat_output.splitlines():
|
||
|
|
match = _RE_STAT_LINE.match(line.decode('utf-8'))
|
||
|
|
if not match:
|
||
|
|
continue
|
||
|
|
key, value = (_remove_redundant_spaces(x) for x in match.groups())
|
||
|
|
attr, convfun = _SOX_STAT_ATTR_MAP.get(key, (None, None))
|
||
|
|
if attr:
|
||
|
|
setattr(stat, attr, convfun(value))
|
||
|
|
|
||
|
|
if not all(hasattr(stat, x[0]) for x in _SOX_STAT_ATTR_MAP.values()):
|
||
|
|
logging.error('stat_output: %s', stat_output)
|
||
|
|
raise RuntimeError('missing entries: ' + str(stat))
|
||
|
|
|
||
|
|
return stat
|
||
|
|
|
||
|
|
|
||
|
|
def convert_raw_file(path_src, channels_src, bits_src, rate_src,
|
||
|
|
path_dst):
|
||
|
|
"""Converts a raw file to a new format.
|
||
|
|
|
||
|
|
@param path_src: The path to the source file.
|
||
|
|
@param channels_src: The channel number of the source file.
|
||
|
|
@param bits_src: The size of sample in bits of the source file.
|
||
|
|
@param rate_src: The sampling rate of the source file.
|
||
|
|
@param path_dst: The path to the destination file. The file name determines
|
||
|
|
the new file format.
|
||
|
|
|
||
|
|
"""
|
||
|
|
sox_cmd = [SOX_PATH]
|
||
|
|
sox_cmd += _raw_format_args(channels_src, bits_src, rate_src)
|
||
|
|
sox_cmd += [path_src]
|
||
|
|
sox_cmd += [path_dst]
|
||
|
|
cmd_utils.execute(sox_cmd)
|
||
|
|
|
||
|
|
|
||
|
|
def convert_format(path_src, channels_src, bits_src, rate_src,
|
||
|
|
path_dst, channels_dst, bits_dst, rate_dst,
|
||
|
|
volume_scale, use_src_header=False, use_dst_header=False):
|
||
|
|
"""Converts a raw file to a new format.
|
||
|
|
|
||
|
|
@param path_src: The path to the source file.
|
||
|
|
@param channels_src: The channel number of the source file.
|
||
|
|
@param bits_src: The size of sample in bits of the source file.
|
||
|
|
@param rate_src: The sampling rate of the source file.
|
||
|
|
@param path_dst: The path to the destination file.
|
||
|
|
@param channels_dst: The channel number of the destination file.
|
||
|
|
@param bits_dst: The size of sample in bits of the destination file.
|
||
|
|
@param rate_dst: The sampling rate of the destination file.
|
||
|
|
@param volume_scale: A float for volume scale used in sox command.
|
||
|
|
E.g. 1.0 is the same. 0.5 to scale volume by
|
||
|
|
half. -1.0 to invert the data.
|
||
|
|
@param use_src_header: True to use header from source file and skip
|
||
|
|
specifying channel, sample format, and rate for
|
||
|
|
source. False otherwise.
|
||
|
|
@param use_dst_header: True to use header for dst file. False to treat
|
||
|
|
dst file as a raw file.
|
||
|
|
|
||
|
|
"""
|
||
|
|
sox_cmd = [SOX_PATH]
|
||
|
|
|
||
|
|
if not use_src_header:
|
||
|
|
sox_cmd += _raw_format_args(channels_src, bits_src, rate_src)
|
||
|
|
sox_cmd += ['-v', '%f' % volume_scale]
|
||
|
|
sox_cmd += [path_src]
|
||
|
|
|
||
|
|
if not use_dst_header:
|
||
|
|
sox_cmd += _raw_format_args(channels_dst, bits_dst, rate_dst)
|
||
|
|
else:
|
||
|
|
sox_cmd += _format_args(channels_dst, bits_dst, rate_dst)
|
||
|
|
sox_cmd += [path_dst]
|
||
|
|
|
||
|
|
cmd_utils.execute(sox_cmd)
|
||
|
|
|
||
|
|
|
||
|
|
def lowpass_filter(path_src, channels_src, bits_src, rate_src,
|
||
|
|
path_dst, frequency):
|
||
|
|
"""Passes a raw file to a lowpass filter.
|
||
|
|
|
||
|
|
@param path_src: The path to the source file.
|
||
|
|
@param channels_src: The channel number of the source file.
|
||
|
|
@param bits_src: The size of sample in bits of the source file.
|
||
|
|
@param rate_src: The sampling rate of the source file.
|
||
|
|
@param path_dst: The path to the destination file.
|
||
|
|
@param frequency: A float for frequency used in sox command. The 3dB
|
||
|
|
frequency of the lowpass filter. Checks manual of sox
|
||
|
|
command for detail.
|
||
|
|
|
||
|
|
"""
|
||
|
|
sox_cmd = [SOX_PATH]
|
||
|
|
sox_cmd += _raw_format_args(channels_src, bits_src, rate_src)
|
||
|
|
sox_cmd += [path_src]
|
||
|
|
sox_cmd += _raw_format_args(channels_src, bits_src, rate_src)
|
||
|
|
sox_cmd += [path_dst]
|
||
|
|
sox_cmd += ['lowpass', '-2', str(frequency)]
|
||
|
|
cmd_utils.execute(sox_cmd)
|
||
|
|
|
||
|
|
|
||
|
|
def trim_silence_from_wav_file(path_src,
|
||
|
|
path_dst,
|
||
|
|
new_duration,
|
||
|
|
volume=1,
|
||
|
|
duration_threshold=0.1):
|
||
|
|
"""Trim silence from beginning of a file.
|
||
|
|
|
||
|
|
Trim silence from beginning of file, and trim remaining audio to
|
||
|
|
new_duration seconds in length.
|
||
|
|
|
||
|
|
@param path_src: The path to the source file.
|
||
|
|
@oaram path_dst: The path to the destination file.
|
||
|
|
@param new_duration: The new duration of the destination file in seconds.
|
||
|
|
@param volume: [Optional] A float indicating the volume in percent, below
|
||
|
|
which sox will consider silence, defaults to 1 (1%).
|
||
|
|
@param duration_threshold: [Optional] A float of the duration in seconds of
|
||
|
|
sound above volume parameter required to consider
|
||
|
|
end of silence. Defaults to 0.1 (0.1 seconds).
|
||
|
|
"""
|
||
|
|
mins, secs = divmod(new_duration, 60)
|
||
|
|
hrs, mins = divmod(mins, 60)
|
||
|
|
length_str = '{:d}:{:02d}:{:.3f}'.format(int(hrs), int(mins), float(secs))
|
||
|
|
|
||
|
|
sox_cmd = [SOX_PATH]
|
||
|
|
sox_cmd += ['-G', path_src, path_dst]
|
||
|
|
sox_cmd += ['silence', '1', str(duration_threshold), '{}%'.format(volume)]
|
||
|
|
sox_cmd += ['trim', '0', length_str]
|
||
|
|
|
||
|
|
cmd_utils.execute(sox_cmd)
|
||
|
|
|
||
|
|
|
||
|
|
def mix_two_wav_files(path_src1, path_src2, path_dst, input_volume=None):
|
||
|
|
"""Generate the mixed WAV file from two input WAV files.
|
||
|
|
|
||
|
|
Use "man sox" for more details on the mixing.
|
||
|
|
|
||
|
|
@param path_src1: Path to the first source.
|
||
|
|
@param path_src2: Path to the second source.
|
||
|
|
@param path_dst: Path for the generated mixed file.
|
||
|
|
@param input_volume: The volume (0.0~1.0) of input sources on mixing. If not
|
||
|
|
given, the default value for sox is 1 / (# of sources).
|
||
|
|
"""
|
||
|
|
sox_cmd = [SOX_PATH]
|
||
|
|
sox_cmd += ['--combine', 'mix']
|
||
|
|
|
||
|
|
if isinstance(input_volume, (int, float)):
|
||
|
|
input_volume = min(1.0, max(0.0, input_volume))
|
||
|
|
sox_cmd += ['-v', '{:.3f}'.format(input_volume)]
|
||
|
|
|
||
|
|
sox_cmd += [path_src1, path_src2, path_dst]
|
||
|
|
|
||
|
|
cmd_utils.execute(sox_cmd)
|
||
|
|
|
||
|
|
|
||
|
|
def get_infos_from_wav_file(file_path):
|
||
|
|
"""Get the information set from the header of the input WAV file.
|
||
|
|
|
||
|
|
It returns None if the input file is not WAV format.
|
||
|
|
|
||
|
|
@param file_path: Path to the WAV file.
|
||
|
|
|
||
|
|
@returns: A dict with the following elements:
|
||
|
|
'duration': The length of the audio (in seconds).
|
||
|
|
'channels': The number of channels.
|
||
|
|
'bits': The number of bits of each sample.
|
||
|
|
'rate': The sampling rate.
|
||
|
|
"""
|
||
|
|
sox_cmd = [SOX_PATH]
|
||
|
|
sox_cmd += ['--i', None, file_path] # sox_cmd[2] is placeholder
|
||
|
|
|
||
|
|
def _execute_sox_cmd_info(info_arg):
|
||
|
|
sox_cmd_info = sox_cmd[:2] + [info_arg] + sox_cmd[3:]
|
||
|
|
return cmd_utils.execute(
|
||
|
|
sox_cmd_info, stdout=subprocess.PIPE).decode('utf-8').strip()
|
||
|
|
|
||
|
|
format_output = _execute_sox_cmd_info('-t')
|
||
|
|
if format_output != 'wav':
|
||
|
|
logging.error('the input file format: %s', format_output)
|
||
|
|
return None
|
||
|
|
|
||
|
|
return dict(duration=float(_execute_sox_cmd_info('-D')),
|
||
|
|
channels=int(_execute_sox_cmd_info('-c')),
|
||
|
|
bits=int(_execute_sox_cmd_info('-b')),
|
||
|
|
rate=int(_execute_sox_cmd_info('-r')))
|
||
|
|
|
||
|
|
|
||
|
|
def get_file_length(file_path, channels, bits, rate):
|
||
|
|
"""Get the length in seconds of an audio file.
|
||
|
|
|
||
|
|
@param file_path: Path to audio file.
|
||
|
|
@param channels: The number of channels.
|
||
|
|
@param bits: The number of bits of each sample.
|
||
|
|
@param rate: The sampling rate.
|
||
|
|
|
||
|
|
@returns: float length in seconds
|
||
|
|
"""
|
||
|
|
return get_stat(file_path, channels, bits, rate).length
|