Main Content

OpenL3 Embeddings

Extract OpenL3 embeddings

Since R2022b

  • OpenL3 Embeddings block

Libraries:
Audio Toolbox / Deep Learning

Description

The OpenL3 Embeddings block uses OpenL3 to extract feature embeddings from audio signals. The OpenL3 Embeddings block combines necessary audio preprocessing and OpenL3 network inference and returns feature embeddings that are a compact representation of audio data. This block requires Deep Learning Toolbox™.

Ports

Input

expand all

Sound data, specified as a one-channel signal (column vector). If Sample rate of input signal (Hz) is 48e3, there are no restrictions on the input frame length. If Sample rate of input signal (Hz) is different from 48e3, then the input frame length must be a multiple of the decimation factor of the resampling operation that the block performs. If the input frame length does not satisfy this condition, the block throws an error message with information on the decimation factor.

Data Types: single | double

Output

expand all

Output embedding, returned as a row vector whose length is specified by the Embedding length parameter.

Data Types: single

Parameters

expand all

Sample rate of the input signal in Hz, specified as a positive scalar.

Specify the overlap percentage between consecutive spectrograms as a scalar in the range [0 100).

Type of spectrum generated from audio and used as input to the neural network, specified as Mel (128 bands), Mel (256 bands), or Linear.

  • Mel (128 bands) –– The neural network accepts mel spectrograms generated from the input audio with 128 mel bands.

  • Mel (256 bands) –– The neural network accepts mel spectrograms generated from the input audio with 256 mel bands.

  • Linear –– The neural network accepts positive one-sided spectrograms generated from the input audio with an FFT length of 257.

Type of audio content the neural network was trained on, specified as Environmental sounds or Musical sounds. Set this parameter to Environmental sounds to use a neural network pretrained on environmental audio data, and set it to Musical sounds to use a network pretrained on musical data.

Length of output embedding, specified as 512 or 6144.

Block Characteristics

Data Types

double | single

Direct Feedthrough

no

Multidimensional Signals

no

Variable-Size Signals

no

Zero-Crossing Detection

no

References

[1] Cramer, Jason, et al. "Look, Listen, and Learn More: Design Choices for Deep Audio Embeddings." In ICASSP 2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), IEEE, 2019, pp. 3852-56. DOI.org (Crossref), doi:/10.1109/ICASSP.2019.8682475.

Extended Capabilities

Version History

Introduced in R2022b