Created
March 19, 2020 05:05
-
-
Save dschwertfeger/f9746bc62871c736e47d5ec3ff4230f7 to your computer and use it in GitHub Desktop.
Convert a power-spectrogram to decibel units in TensorFlow
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
def power_to_db(S, amin=1e-16, top_db=80.0): | |
"""Convert a power-spectrogram (magnitude squared) to decibel (dB) units. | |
Computes the scaling ``10 * log10(S / max(S))`` in a numerically | |
stable way. | |
Based on: | |
https://librosa.github.io/librosa/generated/librosa.core.power_to_db.html | |
""" | |
def _tf_log10(x): | |
numerator = tf.math.log(x) | |
denominator = tf.math.log(tf.constant(10, dtype=numerator.dtype)) | |
return numerator / denominator | |
# Scale magnitude relative to maximum value in S. Zeros in the output | |
# correspond to positions where S == ref. | |
ref = tf.reduce_max(S) | |
log_spec = 10.0 * _tf_log10(tf.maximum(amin, S)) | |
log_spec -= 10.0 * _tf_log10(tf.maximum(amin, ref)) | |
log_spec = tf.maximum(log_spec, tf.reduce_max(log_spec) - top_db) | |
return log_spec |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment