High-Performance Sound Technologies for Access and Scholarship (@hipstas)
Guthub

hipstas

Twitter logo Share LinkedIn logo Share Facebook logo Share

Top repositories

1

audio-labeler

An in-browser app for labeling audio clips at random, using Docker and Flask.
JavaScript
46
star
2

kaldi-pop-up-archive

A Docker image for the Kaldi speech recognition tool + training data from Pop Up Archive
Perl
19
star
3

audio-tagging-toolkit

A Python package for audio annotation and classifier training. Developed in collaboration with the WGBH Foundation and the American Archive of Public Broadcasting.
Python
18
star
4

AudiAnnotate

Workflows for generating AV editions and exhibits using IIIF manifests by HiPSTAS and Brumfield Labs.
Ruby
15
star
5

audio-ml-lab

A Dockerized Jupyter notebook environment with pre-installed audio machine learning tools.
Dockerfile
12
star
6

spokenweb

The development of this workshop for audio analysis has been supported by the SpokenWeb (https://spokenweb.ca/) and the Bridging Barriers Good Systems (https://bridgingbarriers.utexas.edu/good-systems/) projects. Developers for the workshop include Brian McFee, Chris Ick, Liz Fischer, and Tanya Clement.
Jupyter Notebook
6
star
7

sida

Speaker Identification for Archives. This repository includes several notebooks that walks through the steps of training and running a classifier that takes speaker labels and the audio, extracts features (including vowels), and trains a model and runs it.
Jupyter Notebook
4
star
8

documentation

Getting Started
Ruby
3
star
9

aapb-speaker-labels

This repository contains speaker labels in CSV files for training speaker identification classifiers. These speakers appear in a subset of AAPB files.
Jupyter Notebook
3
star
10

aapb-data

Data and code for ongoing collaboration between the High-Performance Sound Technologies for Access and Scholarship research group at UT Austin, the WGBH Foundation, and the American Archive of Public Broadcasting.
Jupyter Notebook
3
star
11

applause-classifier

This repository includes training data and SVM classifier for locating applause in audio recordings.
3
star
12

aapb-ubm

This repository contains preprocessing instructions for building a universal background model for speaker identification in the AAPB corpus.
Jupyter Notebook
2
star
13

american-archive-kaldi

Python
2
star
14

audio-ml-introduction

This repository contains a demonstration workshop run at Indiana University at Bloomington, March 2017.
Jupyter Notebook
2
star
15

test-tone-classifier

A machine learning classifier, including training data, for identifying broadcast test tones in audio and video files.
Jupyter Notebook
1
star
16

audio-ml-lab-server

1
star
17

audio_basic_workflows

This repository contains notebooks with basic and simple workflows for audio processing and analysis in the humanities.
Jupyter Notebook
1
star
18

AudiAnnotateTheme

Jekyll Theme for AudiAnnotate Projects
JavaScript
1
star
19

podcast-speaker-labels

A miscellaneous collection of human-approved audio labels.
Jupyter Notebook
1
star
20

pbcore-mongodb

This repository contains all the pbcore metadata from the AAPB. This includes a script that turns the XML structure into JSON and loads pbcore metadata in mongodb database in order to construct or customize speaker-specific UBMs
Python
1
star
21

aapb-demo-notebooks

These demo notebooks demonstrate how to train and run audio classifiers.
Jupyter Notebook
1
star
22

aapb-classifier-output

Audio classifier output for identifying Marco Werman as a speaker across all the recordings of The World in AAPBin the American Archive of Public Broadcasting. CSV includes start time, duration, confidence level, speaker name
Jupyter Notebook
1
star