Python mel_features.log_mel_spectrogram() Examples

The following are 12 code examples of mel_features.log_mel_spectrogram(). You can vote up the ones you like or vote down the ones you don't like, and go to the original project or source file by following the links above each example. You may also want to check out all available functions/classes of the module mel_features , or try the search function .
Example #1
Source File: vggish_input.py    From edusense with BSD 3-Clause "New" or "Revised" License 5 votes vote down vote up 
def waveform_to_examples(data, sample_rate):
    # Convert to mono.
    if len(data.shape) > 1:
        data = np.mean(data, axis=1)
    # Resample to the rate assumed by VGGish.
    if sample_rate != vggish_params.SAMPLE_RATE:
        data = resampy.resample(data, sample_rate, vggish_params.SAMPLE_RATE)

    # Compute log mel spectrogram features.
    log_mel = mel_features.log_mel_spectrogram(
        data,
        audio_sample_rate=vggish_params.SAMPLE_RATE,
        log_offset=vggish_params.LOG_OFFSET,
        window_length_secs=vggish_params.STFT_WINDOW_LENGTH_SECONDS,
        hop_length_secs=vggish_params.STFT_HOP_LENGTH_SECONDS,
        num_mel_bins=vggish_params.NUM_MEL_BINS,
        lower_edge_hertz=vggish_params.MEL_MIN_HZ,
        upper_edge_hertz=vggish_params.MEL_MAX_HZ)

    # Frame features into examples.
    features_sample_rate = 1.0 / vggish_params.STFT_HOP_LENGTH_SECONDS
    example_window_length = int(round(
        vggish_params.EXAMPLE_WINDOW_SECONDS * features_sample_rate))
    example_hop_length = int(round(
        vggish_params.EXAMPLE_HOP_SECONDS * features_sample_rate))
    log_mel_examples = mel_features.frame(
        log_mel,
        window_length=example_window_length,
        hop_length=example_hop_length)
    return log_mel_examples
Example #2
Source File: model.py    From project-keyword-spotter with Apache License 2.0 5 votes vote down vote up 
def _compute_spectrogram(self, audio_samples, audio_sample_rate_hz):
    """Compute log-mel spectrogram and scale it to uint8."""
    samples = audio_samples.flatten() / float(2**15)
    spectrogram = 30 * (
        mel_features.log_mel_spectrogram(
            samples,
            audio_sample_rate_hz,
            log_offset=0.001,
            window_length_secs=self.spectrogram_window_length_seconds,
            hop_length_secs=self.spectrogram_hop_length_seconds,
            num_mel_bins=self.num_mel_bins,
            lower_edge_hertz=60,
            upper_edge_hertz=3800) - np.log(1e-3))
    return spectrogram
Example #3
Source File: vggish_input.py    From sklearn-audio-transfer-learning with ISC License 4 votes vote down vote up 
def waveform_to_examples(data, sample_rate):
  """Converts audio waveform into an array of examples for VGGish.

  Args:
    data: np.array of either one dimension (mono) or two dimensions
      (multi-channel, with the outer dimension representing channels).
      Each sample is generally expected to lie in the range [-1.0, +1.0],
      although this is not required.
    sample_rate: Sample rate of data.

  Returns:
    3-D np.array of shape [num_examples, num_frames, num_bands] which represents
    a sequence of examples, each of which contains a patch of log mel
    spectrogram, covering num_frames frames of audio and num_bands mel frequency
    bands, where the frame length is vggish_params.STFT_HOP_LENGTH_SECONDS.
  """
  # Convert to mono.
  if len(data.shape) > 1:
    data = np.mean(data, axis=1)
  # Resample to the rate assumed by VGGish.
  if sample_rate != vggish_params.SAMPLE_RATE:
    data = resampy.resample(data, sample_rate, vggish_params.SAMPLE_RATE)

  # Compute log mel spectrogram features.
  log_mel = mel_features.log_mel_spectrogram(
      data,
      audio_sample_rate=vggish_params.SAMPLE_RATE,
      log_offset=vggish_params.LOG_OFFSET,
      window_length_secs=vggish_params.STFT_WINDOW_LENGTH_SECONDS,
      hop_length_secs=vggish_params.STFT_HOP_LENGTH_SECONDS,
      num_mel_bins=vggish_params.NUM_MEL_BINS,
      lower_edge_hertz=vggish_params.MEL_MIN_HZ,
      upper_edge_hertz=vggish_params.MEL_MAX_HZ)

  # Frame features into examples.
  features_sample_rate = 1.0 / vggish_params.STFT_HOP_LENGTH_SECONDS
  example_window_length = int(round(
      vggish_params.EXAMPLE_WINDOW_SECONDS * features_sample_rate))
  example_hop_length = int(round(
      vggish_params.EXAMPLE_HOP_SECONDS * features_sample_rate))
  log_mel_examples = mel_features.frame(
      log_mel,
      window_length=example_window_length,
      hop_length=example_hop_length)
  return log_mel_examples
Example #4
Source File: vggish_input.py    From Tensorflow-Audio-Classification with Apache License 2.0 4 votes vote down vote up 
def waveform_to_examples(data, sample_rate):
  """Converts audio waveform into an array of examples for VGGish.

  Args:
    data: np.array of either one dimension (mono) or two dimensions
      (multi-channel, with the outer dimension representing channels).
      Each sample is generally expected to lie in the range [-1.0, +1.0],
      although this is not required.
    sample_rate: Sample rate of data.

  Returns:
    3-D np.array of shape [num_examples, num_frames, num_bands] which represents
    a sequence of examples, each of which contains a patch of log mel
    spectrogram, covering num_frames frames of audio and num_bands mel frequency
    bands, where the frame length is vggish_params.STFT_HOP_LENGTH_SECONDS.
  """
  # Convert to mono.
  if len(data.shape) > 1:
    data = np.mean(data, axis=1)
  # Resample to the rate assumed by VGGish.
  if sample_rate != vggish_params.SAMPLE_RATE:
    data = resampy.resample(data, sample_rate, vggish_params.SAMPLE_RATE)

  # Compute log mel spectrogram features.
  log_mel = mel_features.log_mel_spectrogram(
      data,
      audio_sample_rate=vggish_params.SAMPLE_RATE,
      log_offset=vggish_params.LOG_OFFSET,
      window_length_secs=vggish_params.STFT_WINDOW_LENGTH_SECONDS,
      hop_length_secs=vggish_params.STFT_HOP_LENGTH_SECONDS,
      num_mel_bins=vggish_params.NUM_MEL_BINS,
      lower_edge_hertz=vggish_params.MEL_MIN_HZ,
      upper_edge_hertz=vggish_params.MEL_MAX_HZ)

  # Frame features into examples.
  features_sample_rate = 1.0 / vggish_params.STFT_HOP_LENGTH_SECONDS
  example_window_length = int(round(
      vggish_params.EXAMPLE_WINDOW_SECONDS * features_sample_rate))
  example_hop_length = int(round(
      vggish_params.EXAMPLE_HOP_SECONDS * features_sample_rate))
  log_mel_examples = mel_features.frame(
      log_mel,
      window_length=example_window_length,
      hop_length=example_hop_length)
  return log_mel_examples
Example #5
Source File: vggish_input.py    From yolo_v2 with Apache License 2.0 4 votes vote down vote up 
def waveform_to_examples(data, sample_rate):
  """Converts audio waveform into an array of examples for VGGish.

  Args:
    data: np.array of either one dimension (mono) or two dimensions
      (multi-channel, with the outer dimension representing channels).
      Each sample is generally expected to lie in the range [-1.0, +1.0],
      although this is not required.
    sample_rate: Sample rate of data.

  Returns:
    3-D np.array of shape [num_examples, num_frames, num_bands] which represents
    a sequence of examples, each of which contains a patch of log mel
    spectrogram, covering num_frames frames of audio and num_bands mel frequency
    bands, where the frame length is vggish_params.STFT_HOP_LENGTH_SECONDS.
  """
  # Convert to mono.
  if len(data.shape) > 1:
    data = np.mean(data, axis=1)
  # Resample to the rate assumed by VGGish.
  if sample_rate != vggish_params.SAMPLE_RATE:
    data = resampy.resample(data, sample_rate, vggish_params.SAMPLE_RATE)

  # Compute log mel spectrogram features.
  log_mel = mel_features.log_mel_spectrogram(
      data,
      audio_sample_rate=vggish_params.SAMPLE_RATE,
      log_offset=vggish_params.LOG_OFFSET,
      window_length_secs=vggish_params.STFT_WINDOW_LENGTH_SECONDS,
      hop_length_secs=vggish_params.STFT_HOP_LENGTH_SECONDS,
      num_mel_bins=vggish_params.NUM_MEL_BINS,
      lower_edge_hertz=vggish_params.MEL_MIN_HZ,
      upper_edge_hertz=vggish_params.MEL_MAX_HZ)

  # Frame features into examples.
  features_sample_rate = 1.0 / vggish_params.STFT_HOP_LENGTH_SECONDS
  example_window_length = int(round(
      vggish_params.EXAMPLE_WINDOW_SECONDS * features_sample_rate))
  example_hop_length = int(round(
      vggish_params.EXAMPLE_HOP_SECONDS * features_sample_rate))
  log_mel_examples = mel_features.frame(
      log_mel,
      window_length=example_window_length,
      hop_length=example_hop_length)
  return log_mel_examples
Example #6
Source File: vggish_input.py    From Gun-Detector with Apache License 2.0 4 votes vote down vote up 
def waveform_to_examples(data, sample_rate):
  """Converts audio waveform into an array of examples for VGGish.

  Args:
    data: np.array of either one dimension (mono) or two dimensions
      (multi-channel, with the outer dimension representing channels).
      Each sample is generally expected to lie in the range [-1.0, +1.0],
      although this is not required.
    sample_rate: Sample rate of data.

  Returns:
    3-D np.array of shape [num_examples, num_frames, num_bands] which represents
    a sequence of examples, each of which contains a patch of log mel
    spectrogram, covering num_frames frames of audio and num_bands mel frequency
    bands, where the frame length is vggish_params.STFT_HOP_LENGTH_SECONDS.
  """
  # Convert to mono.
  if len(data.shape) > 1:
    data = np.mean(data, axis=1)
  # Resample to the rate assumed by VGGish.
  if sample_rate != vggish_params.SAMPLE_RATE:
    data = resampy.resample(data, sample_rate, vggish_params.SAMPLE_RATE)

  # Compute log mel spectrogram features.
  log_mel = mel_features.log_mel_spectrogram(
      data,
      audio_sample_rate=vggish_params.SAMPLE_RATE,
      log_offset=vggish_params.LOG_OFFSET,
      window_length_secs=vggish_params.STFT_WINDOW_LENGTH_SECONDS,
      hop_length_secs=vggish_params.STFT_HOP_LENGTH_SECONDS,
      num_mel_bins=vggish_params.NUM_MEL_BINS,
      lower_edge_hertz=vggish_params.MEL_MIN_HZ,
      upper_edge_hertz=vggish_params.MEL_MAX_HZ)

  # Frame features into examples.
  features_sample_rate = 1.0 / vggish_params.STFT_HOP_LENGTH_SECONDS
  example_window_length = int(round(
      vggish_params.EXAMPLE_WINDOW_SECONDS * features_sample_rate))
  example_hop_length = int(round(
      vggish_params.EXAMPLE_HOP_SECONDS * features_sample_rate))
  log_mel_examples = mel_features.frame(
      log_mel,
      window_length=example_window_length,
      hop_length=example_hop_length)
  return log_mel_examples
Example #7
Source File: vggish_input.py    From object_detection_kitti with Apache License 2.0 4 votes vote down vote up 
def waveform_to_examples(data, sample_rate):
  """Converts audio waveform into an array of examples for VGGish.

  Args:
    data: np.array of either one dimension (mono) or two dimensions
      (multi-channel, with the outer dimension representing channels).
      Each sample is generally expected to lie in the range [-1.0, +1.0],
      although this is not required.
    sample_rate: Sample rate of data.

  Returns:
    3-D np.array of shape [num_examples, num_frames, num_bands] which represents
    a sequence of examples, each of which contains a patch of log mel
    spectrogram, covering num_frames frames of audio and num_bands mel frequency
    bands, where the frame length is vggish_params.STFT_HOP_LENGTH_SECONDS.
  """
  # Convert to mono.
  if len(data.shape) > 1:
    data = np.mean(data, axis=1)
  # Resample to the rate assumed by VGGish.
  if sample_rate != vggish_params.SAMPLE_RATE:
    data = resampy.resample(data, sample_rate, vggish_params.SAMPLE_RATE)

  # Compute log mel spectrogram features.
  log_mel = mel_features.log_mel_spectrogram(
      data,
      audio_sample_rate=vggish_params.SAMPLE_RATE,
      log_offset=vggish_params.LOG_OFFSET,
      window_length_secs=vggish_params.STFT_WINDOW_LENGTH_SECONDS,
      hop_length_secs=vggish_params.STFT_HOP_LENGTH_SECONDS,
      num_mel_bins=vggish_params.NUM_MEL_BINS,
      lower_edge_hertz=vggish_params.MEL_MIN_HZ,
      upper_edge_hertz=vggish_params.MEL_MAX_HZ)

  # Frame features into examples.
  features_sample_rate = 1.0 / vggish_params.STFT_HOP_LENGTH_SECONDS
  example_window_length = int(round(
      vggish_params.EXAMPLE_WINDOW_SECONDS * features_sample_rate))
  example_hop_length = int(round(
      vggish_params.EXAMPLE_HOP_SECONDS * features_sample_rate))
  log_mel_examples = mel_features.frame(
      log_mel,
      window_length=example_window_length,
      hop_length=example_hop_length)
  return log_mel_examples
Example #8
Source File: vggish_input.py    From object_detection_with_tensorflow with MIT License 4 votes vote down vote up 
def waveform_to_examples(data, sample_rate):
  """Converts audio waveform into an array of examples for VGGish.

  Args:
    data: np.array of either one dimension (mono) or two dimensions
      (multi-channel, with the outer dimension representing channels).
      Each sample is generally expected to lie in the range [-1.0, +1.0],
      although this is not required.
    sample_rate: Sample rate of data.

  Returns:
    3-D np.array of shape [num_examples, num_frames, num_bands] which represents
    a sequence of examples, each of which contains a patch of log mel
    spectrogram, covering num_frames frames of audio and num_bands mel frequency
    bands, where the frame length is vggish_params.STFT_HOP_LENGTH_SECONDS.
  """
  # Convert to mono.
  if len(data.shape) > 1:
    data = np.mean(data, axis=1)
  # Resample to the rate assumed by VGGish.
  if sample_rate != vggish_params.SAMPLE_RATE:
    data = resampy.resample(data, sample_rate, vggish_params.SAMPLE_RATE)

  # Compute log mel spectrogram features.
  log_mel = mel_features.log_mel_spectrogram(
      data,
      audio_sample_rate=vggish_params.SAMPLE_RATE,
      log_offset=vggish_params.LOG_OFFSET,
      window_length_secs=vggish_params.STFT_WINDOW_LENGTH_SECONDS,
      hop_length_secs=vggish_params.STFT_HOP_LENGTH_SECONDS,
      num_mel_bins=vggish_params.NUM_MEL_BINS,
      lower_edge_hertz=vggish_params.MEL_MIN_HZ,
      upper_edge_hertz=vggish_params.MEL_MAX_HZ)

  # Frame features into examples.
  features_sample_rate = 1.0 / vggish_params.STFT_HOP_LENGTH_SECONDS
  example_window_length = int(round(
      vggish_params.EXAMPLE_WINDOW_SECONDS * features_sample_rate))
  example_hop_length = int(round(
      vggish_params.EXAMPLE_HOP_SECONDS * features_sample_rate))
  log_mel_examples = mel_features.frame(
      log_mel,
      window_length=example_window_length,
      hop_length=example_hop_length)
  return log_mel_examples
Example #9
Source File: vggish_input.py    From audioset_classification with MIT License 4 votes vote down vote up 
def waveform_to_examples(data, sample_rate):
  """Converts audio waveform into an array of examples for VGGish.

  Args:
    data: np.array of either one dimension (mono) or two dimensions
      (multi-channel, with the outer dimension representing channels).
      Each sample is generally expected to lie in the range [-1.0, +1.0],
      although this is not required.
    sample_rate: Sample rate of data.

  Returns:
    3-D np.array of shape [num_examples, num_frames, num_bands] which represents
    a sequence of examples, each of which contains a patch of log mel
    spectrogram, covering num_frames frames of audio and num_bands mel frequency
    bands, where the frame length is vggish_params.STFT_HOP_LENGTH_SECONDS.
  """
  # Convert to mono.
  if len(data.shape) > 1:
    data = np.mean(data, axis=1)
  # Resample to the rate assumed by VGGish.
  if sample_rate != vggish_params.SAMPLE_RATE:
    data = resampy.resample(data, sample_rate, vggish_params.SAMPLE_RATE)

  # Compute log mel spectrogram features.
  log_mel = mel_features.log_mel_spectrogram(
      data,
      audio_sample_rate=vggish_params.SAMPLE_RATE,
      log_offset=vggish_params.LOG_OFFSET,
      window_length_secs=vggish_params.STFT_WINDOW_LENGTH_SECONDS,
      hop_length_secs=vggish_params.STFT_HOP_LENGTH_SECONDS,
      num_mel_bins=vggish_params.NUM_MEL_BINS,
      lower_edge_hertz=vggish_params.MEL_MIN_HZ,
      upper_edge_hertz=vggish_params.MEL_MAX_HZ)

  # Frame features into examples.
  features_sample_rate = 1.0 / vggish_params.STFT_HOP_LENGTH_SECONDS
  example_window_length = int(round(
      vggish_params.EXAMPLE_WINDOW_SECONDS * features_sample_rate))
  example_hop_length = int(round(
      vggish_params.EXAMPLE_HOP_SECONDS * features_sample_rate))
  log_mel_examples = mel_features.frame(
      log_mel,
      window_length=example_window_length,
      hop_length=example_hop_length)
  return log_mel_examples
Example #10
Source File: vggish_input.py    From g-tensorflow-models with Apache License 2.0 4 votes vote down vote up 
def waveform_to_examples(data, sample_rate):
  """Converts audio waveform into an array of examples for VGGish.

  Args:
    data: np.array of either one dimension (mono) or two dimensions
      (multi-channel, with the outer dimension representing channels).
      Each sample is generally expected to lie in the range [-1.0, +1.0],
      although this is not required.
    sample_rate: Sample rate of data.

  Returns:
    3-D np.array of shape [num_examples, num_frames, num_bands] which represents
    a sequence of examples, each of which contains a patch of log mel
    spectrogram, covering num_frames frames of audio and num_bands mel frequency
    bands, where the frame length is vggish_params.STFT_HOP_LENGTH_SECONDS.
  """
  # Convert to mono.
  if len(data.shape) > 1:
    data = np.mean(data, axis=1)
  # Resample to the rate assumed by VGGish.
  if sample_rate != vggish_params.SAMPLE_RATE:
    data = resampy.resample(data, sample_rate, vggish_params.SAMPLE_RATE)

  # Compute log mel spectrogram features.
  log_mel = mel_features.log_mel_spectrogram(
      data,
      audio_sample_rate=vggish_params.SAMPLE_RATE,
      log_offset=vggish_params.LOG_OFFSET,
      window_length_secs=vggish_params.STFT_WINDOW_LENGTH_SECONDS,
      hop_length_secs=vggish_params.STFT_HOP_LENGTH_SECONDS,
      num_mel_bins=vggish_params.NUM_MEL_BINS,
      lower_edge_hertz=vggish_params.MEL_MIN_HZ,
      upper_edge_hertz=vggish_params.MEL_MAX_HZ)

  # Frame features into examples.
  features_sample_rate = 1.0 / vggish_params.STFT_HOP_LENGTH_SECONDS
  example_window_length = int(round(
      vggish_params.EXAMPLE_WINDOW_SECONDS * features_sample_rate))
  example_hop_length = int(round(
      vggish_params.EXAMPLE_HOP_SECONDS * features_sample_rate))
  log_mel_examples = mel_features.frame(
      log_mel,
      window_length=example_window_length,
      hop_length=example_hop_length)
  return log_mel_examples
Example #11
Source File: vggish_input.py    From models with Apache License 2.0 4 votes vote down vote up 
def waveform_to_examples(data, sample_rate):
  """Converts audio waveform into an array of examples for VGGish.

  Args:
    data: np.array of either one dimension (mono) or two dimensions
      (multi-channel, with the outer dimension representing channels).
      Each sample is generally expected to lie in the range [-1.0, +1.0],
      although this is not required.
    sample_rate: Sample rate of data.

  Returns:
    3-D np.array of shape [num_examples, num_frames, num_bands] which represents
    a sequence of examples, each of which contains a patch of log mel
    spectrogram, covering num_frames frames of audio and num_bands mel frequency
    bands, where the frame length is vggish_params.STFT_HOP_LENGTH_SECONDS.
  """
  # Convert to mono.
  if len(data.shape) > 1:
    data = np.mean(data, axis=1)
  # Resample to the rate assumed by VGGish.
  if sample_rate != vggish_params.SAMPLE_RATE:
    data = resampy.resample(data, sample_rate, vggish_params.SAMPLE_RATE)

  # Compute log mel spectrogram features.
  log_mel = mel_features.log_mel_spectrogram(
      data,
      audio_sample_rate=vggish_params.SAMPLE_RATE,
      log_offset=vggish_params.LOG_OFFSET,
      window_length_secs=vggish_params.STFT_WINDOW_LENGTH_SECONDS,
      hop_length_secs=vggish_params.STFT_HOP_LENGTH_SECONDS,
      num_mel_bins=vggish_params.NUM_MEL_BINS,
      lower_edge_hertz=vggish_params.MEL_MIN_HZ,
      upper_edge_hertz=vggish_params.MEL_MAX_HZ)

  # Frame features into examples.
  features_sample_rate = 1.0 / vggish_params.STFT_HOP_LENGTH_SECONDS
  example_window_length = int(round(
      vggish_params.EXAMPLE_WINDOW_SECONDS * features_sample_rate))
  example_hop_length = int(round(
      vggish_params.EXAMPLE_HOP_SECONDS * features_sample_rate))
  log_mel_examples = mel_features.frame(
      log_mel,
      window_length=example_window_length,
      hop_length=example_hop_length)
  return log_mel_examples
Example #12
Source File: vggish_input.py    From multilabel-image-classification-tensorflow with MIT License 4 votes vote down vote up 
def waveform_to_examples(data, sample_rate):
  """Converts audio waveform into an array of examples for VGGish.

  Args:
    data: np.array of either one dimension (mono) or two dimensions
      (multi-channel, with the outer dimension representing channels).
      Each sample is generally expected to lie in the range [-1.0, +1.0],
      although this is not required.
    sample_rate: Sample rate of data.

  Returns:
    3-D np.array of shape [num_examples, num_frames, num_bands] which represents
    a sequence of examples, each of which contains a patch of log mel
    spectrogram, covering num_frames frames of audio and num_bands mel frequency
    bands, where the frame length is vggish_params.STFT_HOP_LENGTH_SECONDS.
  """
  # Convert to mono.
  if len(data.shape) > 1:
    data = np.mean(data, axis=1)
  # Resample to the rate assumed by VGGish.
  if sample_rate != vggish_params.SAMPLE_RATE:
    data = resampy.resample(data, sample_rate, vggish_params.SAMPLE_RATE)

  # Compute log mel spectrogram features.
  log_mel = mel_features.log_mel_spectrogram(
      data,
      audio_sample_rate=vggish_params.SAMPLE_RATE,
      log_offset=vggish_params.LOG_OFFSET,
      window_length_secs=vggish_params.STFT_WINDOW_LENGTH_SECONDS,
      hop_length_secs=vggish_params.STFT_HOP_LENGTH_SECONDS,
      num_mel_bins=vggish_params.NUM_MEL_BINS,
      lower_edge_hertz=vggish_params.MEL_MIN_HZ,
      upper_edge_hertz=vggish_params.MEL_MAX_HZ)

  # Frame features into examples.
  features_sample_rate = 1.0 / vggish_params.STFT_HOP_LENGTH_SECONDS
  example_window_length = int(round(
      vggish_params.EXAMPLE_WINDOW_SECONDS * features_sample_rate))
  example_hop_length = int(round(
      vggish_params.EXAMPLE_HOP_SECONDS * features_sample_rate))
  log_mel_examples = mel_features.frame(
      log_mel,
      window_length=example_window_length,
      hop_length=example_hop_length)
  return log_mel_examples