B )`W% @sdZddlmZddlmZddlmZddlmZddlmZddlmZddl m Z dd l m Z dd l m Z dd lmZdd lmZd ZdZdddZdddZddZedejdddddejdfddZdS) zmel conversion ops.)absolute_import)division)print_function)dtypes)ops) tensor_util) array_ops)math_ops) shape_ops)dispatch) tf_exportg@g@Nc Cs<t|d|g"t|}tt|tdSQRXdS)a6Converts frequencies in `mel_values` from the mel scale to linear scale. Args: mel_values: A `Tensor` of frequencies in the mel scale. name: An optional name for the operation. Returns: A `Tensor` of the same shape and type as `mel_values` containing linear scale frequencies in Hertz. Z mel_to_hertzg?N)r name_scopeconvert_to_tensor_MEL_BREAK_FREQUENCY_HERTZr exp_MEL_HIGH_FREQUENCY_Q)Z mel_valuesnamerS/home/dcms/DCMS/lib/python3.7/site-packages/tensorflow/python/ops/signal/mel_ops.py _mel_to_hertz$s  rc Cs<t|d|g"t|}ttd|tSQRXdS)a5Converts frequencies in `frequencies_hertz` in Hertz to the mel scale. Args: frequencies_hertz: A `Tensor` of frequencies in Hertz. name: An optional name for the operation. Returns: A `Tensor` of the same shape and type of `frequencies_hertz` containing frequencies in the mel scale. Z hertz_to_melg?N)rr rrr logr)Zfrequencies_hertzrrrr _hertz_to_mel6s  rcCs|dkrtd||dkr(td|||kr@td||ft|tjs||dkr`td|||dkr|td||f|jstd |d S) z1Checks the inputs to linear_to_mel_weight_matrix.rz&num_mel_bins must be positive. Got: %sgz.lower_edge_hertz must be non-negative. Got: %sz.lower_edge_hertz %.1f >= upper_edge_hertz %.1fz%sample_rate must be positive. Got: %szlupper_edge_hertz must not be larger than the Nyquist frequency (sample_rate / 2). Got %s for sample_rate: %sz,dtype must be a floating point type. Got: %sN) ValueError isinstancerTensorZ is_floating) num_mel_bins sample_ratelower_edge_hertzupper_edge_hertzdtyperrr_validate_argumentsGs       r!z"signal.linear_to_mel_weight_matrixi@g@_@g@c sXt|d@}t|tjr2t|}|dk r2|}t||||tj||dd}tj ||dd}tj ||dd}t d|}d} |d } t || || d} t t | d} tjt t |t |d d dd } tfd dt j| d ddD\}}}| |||}|| ||}t|t||}t j|| dgddgg|dSQRXdS)a Returns a matrix to warp linear scale spectrograms to the [mel scale][mel]. Returns a weight matrix that can be used to re-weight a `Tensor` containing `num_spectrogram_bins` linearly sampled frequency information from `[0, sample_rate / 2]` into `num_mel_bins` frequency information from `[lower_edge_hertz, upper_edge_hertz]` on the [mel scale][mel]. This function follows the [Hidden Markov Model Toolkit (HTK)](http://htk.eng.cam.ac.uk/) convention, defining the mel scale in terms of a frequency in hertz according to the following formula: $$ extrm{mel}(f) = 2595 * extrm{log}_{10}(1 + rac{f}{700})$$ In the returned matrix, all the triangles (filterbanks) have a peak value of 1.0. For example, the returned matrix `A` can be used to right-multiply a spectrogram `S` of shape `[frames, num_spectrogram_bins]` of linear scale spectrum values (e.g. STFT magnitudes) to generate a "mel spectrogram" `M` of shape `[frames, num_mel_bins]`. # `S` has shape [frames, num_spectrogram_bins] # `M` has shape [frames, num_mel_bins] M = tf.matmul(S, A) The matrix can be used with `tf.tensordot` to convert an arbitrary rank `Tensor` of linear-scale spectral bins into the mel scale. # S has shape [..., num_spectrogram_bins]. # M has shape [..., num_mel_bins]. M = tf.tensordot(S, A, 1) Args: num_mel_bins: Python int. How many bands in the resulting mel spectrum. num_spectrogram_bins: An integer `Tensor`. How many bins there are in the source spectrogram data, which is understood to be `fft_size // 2 + 1`, i.e. the spectrogram only contains the nonredundant FFT bins. sample_rate: An integer or float `Tensor`. Samples per second of the input signal used to create the spectrogram. Used to figure out the frequencies corresponding to each spectrogram bin, which dictates how they are mapped into the mel scale. lower_edge_hertz: Python float. Lower bound on the frequencies to be included in the mel spectrum. This corresponds to the lower edge of the lowest triangular band. upper_edge_hertz: Python float. The desired top edge of the highest frequency band. dtype: The `DType` of the result matrix. Must be a floating point type. name: An optional name for the operation. Returns: A `Tensor` of shape `[num_spectrogram_bins, num_mel_bins]`. Raises: ValueError: If `num_mel_bins`/`num_spectrogram_bins`/`sample_rate` are not positive, `lower_edge_hertz` is negative, frequency edges are incorrectly ordered, `upper_edge_hertz` is larger than the Nyquist frequency. [mel]: https://en.wikipedia.org/wiki/Mel_scale linear_to_mel_weight_matrixNr)rrrgg@r)Z frame_lengthZ frame_stepc3s|]}t|dgVqdS)r%N)rZreshape).0t)rrr sz.linear_to_mel_weight_matrix..)Zaxisr)rr rrrZconstant_valuer!r castrZlinspacerZ expand_dimsrr frametuplesplitmaximumZminimumpad)rZnum_spectrogram_binsrrrr rZmaybe_const_valZzeroZ bands_to_zeroZ nyquist_hertzZlinear_frequenciesZspectrogram_bins_melZband_edges_melZlower_edge_melZ center_melZupper_edge_melZ lower_slopesZ upper_slopesZmel_weights_matrixr)rrr$]sFD            r$)N)N)__doc__ __future__rrrZtensorflow.python.frameworkrrrZtensorflow.python.opsrr Ztensorflow.python.ops.signalr Ztensorflow.python.utilr Z tensorflow.python.util.tf_exportr rrrrr!Zadd_dispatch_supportfloat32r$rrrrs2