Args: inputs: A 3-D tensor with shape of [batch, time, depth]. filters: An int. Number of outputs (=activation maps) size: An i

Denormalize the normalized values to be original dB-scaled value. :param norm_db: Normalized spectrogram. :param max_db: Maximum dB.

Invert the pre-emphasized waveform to the original waveform. Parameters ---------- preem_wav : np.ndarray [shape=(n,)] The p

Applies {batch|layer} normalization. Args: inputs: A tensor with 2 or more dimensions, where the first dimension has `batch_siz

Emphasize high frequency range of the waveform by increasing power(squared amplitude). Parameters ---------- wav : np.ndarray [shape

Prenet for Encoder and Decoder. Args: inputs: A 3D tensor of shape [N, T, hp.embed_size]. is_training: A boolean. scope: Optiona

Get a waveform from the magnitude spectrogram by Griffin-Lim Algorithm. Parameters ---------- mag : np.ndarray [shape=(1 + n_fft/2,

Applies a series of conv1d separately. Args: inputs: A 3d tensor with shape of [N, T, C] K: An int. The size of conv1d banks. Tha

This is applied in `train1` or `test1` phase.

This is applied in `train2`, `test2` or `convert` phase.

Applies a GRU. Args: inputs: A 3d tensor with shape of [N, T, C]. num_units: An int. The number of hidden units. bidirectio

Highway networks, see https://arxiv.org/abs/1505.00387 Args: inputs: A 3D tensor of shape [N, T, W]. num_units: An int or `None`. Spe

Convert a linear-spectrogram to mel-spectrogram. :param linear: Linear-spectrogram. :param sr: Sample rate. :param n_fft: FFT window

Normalize dB-scaled spectrogram values to be in range of 0~1. :param db: Decibel-scaled spectrogram. :param max_db: Maximum dB. :para

Parameters: correct_labels : These are your true classification categories. predict_labels : T

Get magnitude and phase spectrogram from waveforms. Parameters ---------- wav : np.ndarray [shape=(n,)] The real-valued wave

Args: var_regex (string): regular expression to match variable to update. verbose (bool): whether to print warning ab

Split path to basename, filename and extension. For example, 'a/b/c.wav' => ('a/b', 'c', 'wav') :param path: file path :return: basename,

Applies a GRU to `inputs`, while attending `memory`. Args: inputs: A 3d tensor with shape of [N, T', C']. Decoder inputs. memory: A 3d

Randomly cropped a part in a wav file. :param wav: a waveform :param length: length to be randomly cropped. :return: a randomly cropp

Execute dynamic range compression(https://en.wikipedia.org/wiki/Dynamic_range_compression) to dB. :param db: Decibel-scaled magnitudes :p

Embeds a given tensor. Args: inputs: A `Tensor` with type `int32` or `int64` containing the ids to be looked up in `lookup ta

Emphasize a magnitude spectrogram by applying power function. This is used for removing noise. :param mag: magnitude spectrogram. :param

Read mp3 file from source path, convert it to wav and write it to target path. Necessary libraries: ffmpeg, libav. :param src_path: sou

Read a wav, change sample rate, format, and average decibel and write to target path. :param source_path: source wav file path :param tar

'a/b/c.wav' => ('a/b', 'c', 'wav') :param path: filepath = 'a/b/c.wav' :return: basename, filename, and extension = ('a/b', 'c', 'wav')