Pretrained Models

Transfer learning, sound classification, feature embeddings, pretrained audio deep learning networks

Audio Toolbox™ provides MATLAB^® and Simulink^® support for pretrained audio deep learning networks. Locate and classify sounds with YAMNet and estimate pitch with CREPE. Extract VGGish or OpenL3 feature embeddings to input to machine learning and deep learning systems. Use i-vector systems to produce compact representations of audio signals for applications such as speaker recognition, verification, identification, and diarization, speech emotion recognition, and acoustic machine fault detection.

This functionality requires Deep Learning Toolbox™. The Audio Toolbox pretrained networks are available in Deep Network Designer (Deep Learning Toolbox).

Functions

expand all

VGGish

`vggish`	VGGish neural network
`vggishPreprocess`	Preprocess audio for VGGish feature extraction
`vggishEmbeddings`	Extract VGGish feature embeddings

YAMNet

`classifySound`	Classify sounds in audio signal
`yamnet`	YAMNet neural network
`yamnetGraph`	Graph of YAMNet AudioSet ontology
`yamnetPreprocess`	Preprocess audio for YAMNet classification

OpenL3

`openl3`	OpenL3 neural network
`openl3Preprocess`	Preprocess audio for OpenL3 feature extraction
`openl3Embeddings`	Extract OpenL3 feature embeddings

CREPE

`crepe`	CREPE neural network
`crepePreprocess`	Preprocess audio for CREPE deep learning network
`crepePostprocess`	Postprocess output of CREPE deep learning network
`pitchnn`	Estimate pitch with deep learning neural network

i-Vectors

`ivectorSystem`	Create i-vector system
`speakerRecognition`	Pretrained speaker recognition system

Blocks

expand all

VGGish

VGGish Embeddings	Extract VGGish embeddings
VGGish Preprocess	Preprocess audio for VGGish feature extraction
VGGish	VGGish embeddings extraction network

YAMNet

Sound Classifier	Classify sounds in audio signal
YAMNet	YAMNet sound classification network
YAMNet Preprocess	Preprocess audio for YAMNet classification

OpenL3

OpenL3 Embeddings	Extract OpenL3 embeddings
OpenL3 Preprocess	Preprocess audio for OpenL3 embeddings extraction
OpenL3	OpenL3 embeddings extraction network

Apps

Deep Network Designer

Design, visualize, and train deep learning networks

Topics

Audio Transfer Learning Using Experiment Manager
Configure an experiment that compares the performance of multiple pretrained networks applied to a speech command recognition task using transfer learning.

Related Information

Featured Examples

Speech Command Recognition Using Deep Learning

Use a pretrained deep learning model to perform speech command recognition on streaming audio.

Open Live Script

Transfer Learning with Pretrained Audio Networks in Deep Network Designer

Interactively fine-tune a pretrained network to classify new audio signals using Deep Network Designer.

Open Live Script

3-D Sound Event Localization and Detection Using Trained Recurrent Convolutional Neural Network

Perform 3-D sound event localization and detection using a pretrained deep learning model.

Open Live Script

3-D Speech Enhancement Using Trained Filter and Sum Network

Perform speech enhancement using a pretrained filter and sum network (FaSNet) with ambisonic data.

Open Live Script

Speech Command Recognition in Simulink

Detect the presence of speech commands in audio using a Simulink model.

Open Model

Keyword Spotting in Simulink

Use a pretrained deep learning model in Simulink to identify a keyword in speech.

Open Live Script

Investigate Audio Classifications Using Deep Learning Interpretability Techniques

Use interpretability techniques to investigate the predictions of a deep neural network trained to classify audio data.

Open Live Script