onramplab/laravel-transcription

A Laravel package providing transcription feature for audio file

v0.1.0 2023-09-07 11:41 UTC

Requires

• php: >=8.1
• aws/aws-sdk-php: ^3.269
• guzzlehttp/guzzle: ^6.5
• illuminate/database: ^8.0|^9.0
• php-ffmpeg/php-ffmpeg: ^1.1

Requires (Dev)

• doctrine/dbal: ^3.6
• mockery/mockery: ^1.5
• nunomaduro/larastan: ^1.0
• nunomaduro/phpinsights: ^2.6
• onramplab/onr-phpcs-laravel: ^1.2
• orchestra/testbench: ^6.25
• phpmd/phpmd: ^2.13
• phpunit/phpunit: ^9.5
• rector/rector: ^0.15.3
• sempro/phpunit-pretty-print: ^1.4
• spatie/phpunit-watcher: ^1.23
• squizlabs/php_codesniffer: *

MIT

• Onramp Lab

• audio processing
• laravel
• Transcription
• transcript
• speech to text

README

A Laravel package providing transcription feature for audio file

Requirements

• PHP >= 8.1;
• composer.

Features

• Support multiple types of third-party transcription service
  • AWS Transcribe

Installation

composer require onramplab/laravel-transcription

Publish migration files and run command to build tables needed in package

php artisan vendor:publish --tag="transcription-migrations"
php artisan migrate

Also, you can choose to publish the configuration file

php artisan vendor:publish --tag="transcription-config"

Usage

Transcription

1. Set up credentials for audio transcriber you want to use for transcription in your config/transcription.php configuration file.
2. Use Transcription facade to start a transcription.

Transcription::make('https://www.example.com/audio/test.wav', 'en-US');

The Transcription facade's make method is used to start a asynchronous transcription process. The first argument is the URL that can located the audio file you want to transcribe, and the second argument is the language of the supplied audio as a BCP-47 language tag.

Speaker Identification

Sometimes you might want to identify the speaker in your audio file, you can pass the maximum of speaker count as third argument of Transcription facade's make method

Transcription::make('https://www.example.com/audio/test.wav', 'en-US', 3);

After transcription of audio file completed, you can check out the speaker_label attribute of each TranscriptSegment models related with the Transcript model. There would be identifiable information of the speaker for each transcript segments. For example, you might see speaker_1 in that column to represent the first speaker.

Redaction

For security propose, you might need to hide PII (Personal Identifiable Information) data in transcript. Then you should set up credentials for detector you want to use for PII entity detection in your config/transcription.php configuration file.

After setting up configuration file, you can use the fourth argument of Transcription facade's make method to decide whether the transcript of audio file should be redacted.

Transcription::make('https://www.example.com/audio/test.wav', 'en-US', shouldRedact: true);

Useful Tools

Running Tests:

php vendor/bin/phpunit

or

composer test

Code Sniffer Tool:

php vendor/bin/phpcs --standard=PSR2 src/

or

composer psr2check

Code Auto-fixer:

composer psr2autofix
composer insights:fix
rector:fix

Building Docs:

php vendor/bin/phpdoc -d "src" -t "docs"

or

composer docs

Changelog

To keep track, please refer to CHANGELOG.md.

Contributing

1. Fork it.
2. Create your feature branch (git checkout -b my-new-feature).
3. Make your changes.
4. Run the tests, adding new ones for your own code if necessary (phpunit).
5. Commit your changes (git commit -am 'Added some feature').
6. Push to the branch (git push origin my-new-feature).
7. Create new pull request.

Also please refer to CONTRIBUTION.md.

License

Please refer to LICENSE.