An easy to use package for parsing media and transforming it for your Machine Learning projects.
Project description
ML Formatter
A simple set of scripts that take act as a middle man for media intake during machine learning pipelines.
Given input arguments, this program aims to produce the required output format for specific use cases.
Currently:
- DeepSpeech
Usage
> python -m formatter
To view all arguments
> python -m formatter --help
usage: __main__.py [-h] [-v | -q] [--dont-shuffle] [--train TRAIN] [--test TEST] [--val VAL] [--parser {deepspeech}] [--media_type {wav}]
[--transcript_type {txt}] [--media MEDIA] [--transcript TRANSCRIPT] [--output OUTPUT]
Given a set of media, create and output to the required spec of certain ML programs
optional arguments:
-h, --help show this help message and exit
-v, --verbose
-q, --quiet
--dont-shuffle Don't shuffle before splitting into runs
--train TRAIN Training part of train/test/val split. Out of 1
--test TEST Testing part of train/test/val split. Out of 1
--val VAL Validation part of train/test/val split. Out of 1
--parser {deepspeech}
The format you wish to receive as output
--media_type {wav} The file extension of media files
--transcript_type {txt}
The file extension of text transcript files
--media MEDIA Path to the directory containing media files
--transcript TRANSCRIPT
Path to files containing text transcripts
--output OUTPUT Path to directory to use as an output folder
Example usage
> python -m formatter --media ./media --transcript ./transcripts --output ./output --verbose
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
ML-Formatter-1.0.0.tar.gz
(6.2 kB
view hashes)
Built Distribution
Close
Hashes for ML_Formatter-1.0.0-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | d5a0eee33a79f7cd809608034f860680e3ec24fe3bb06c0a77f0677b6cf396dd |
|
MD5 | e77c8ba45397ff9e7f57182acedc46cf |
|
BLAKE2b-256 | 5f95d99db99ce779fd05110b509363c61c91b459bcdefa302aa74bc5955c4aa1 |