README

This Laravel package provides a convenient interface for the Google Speech to Text API.

Prerequisites

The gRPC packages is required when enabling the word time offsets option
- Step 1: Run pecl install grpc
  - gRPC PHP Quick Start
- Step 2: Add extension=grpc.so to php.ini
  - grpc.dll on windows

Installation

composer require noardcode/speech-to-text

Getting started

Open Google Cloud Console and add the Cloud Speech-to-Text API to API's en services in your project.
Create an Google Service Account with the following role: Cloud Speech Service Agent
- Make sure to generate a Service Account Key this file will be used for Authentication.
Run php artisan vendor:publish --provider="Noardcode\SpeechToText\SpeechToTextServiceProvider"
- This will create a speech-to-text.php file in your config folder.
In speech-to-text.php change the following

/*
|--------------------------------------------------------------------------
| Google Service Account
|--------------------------------------------------------------------------
*/
'service-account' => '/path/to/service-account.json',

For a detailed documentation about service accounts see: https://cloud.google.com/docs/authentication/production

Basic examples

// Run on Google Cloud Storage object
resolve(SpeechToText::class)->run('gs://your-bucket-name/path-to-object');

// Run on stored audio file (needs to be: less than 10MB in size and less than 1 minute in length)
resolve(SpeechToText::class)
    ->setAudio(new FilesystemAudio)
    ->run('/path/to/audio-file');

// Using different types of transcripts (e.g. include word time offsets (startTime and endTime))
resolve(SpeechToText::class)->run('gs://your-bucket-name/path-to-object')
    ->setTranscript(new WordTimeOffsets)
    ->run('gs://your-bucket-name/path-to-object');

Settings

You can change the default settings by publishing the config file and changing the following values.

/*
|--------------------------------------------------------------------------
| Default parameters injected by the Service Provider
|--------------------------------------------------------------------------
*/
'defaults' => [
    'language' => 'en-US',
    'encoding' => \Google\Cloud\Speech\V1\RecognitionConfig\AudioEncoding::LINEAR16,
    'sampleRateHertz' => 44100
]

Or change the settings when you have an instance of the class.

$speechToText = resolve(SpeechToText::class)
    ->setLanguageCode('en-US')
    ->setEncoding(\Google\Cloud\Speech\V1\RecognitionConfig\AudioEncoding::LINEAR16)
    ->setSampleRateHertz(44100);

Audio types

By default the SpeechToText class will be passed the a GoogleCloudStorageAudio class. This class tells the SpeechToText class how to create the RecognitionAudio class from the Google Speech to Text Package. If you want to create the RecognitionAudio in in different way, e.g. a file from your local filesystem, you will need to set an other Audio class that implements the AudioInterface.

// Run on audio file on local filesyem 
resolve(SpeechToText::class)
    ->setAudio(new FilesystemAudio)
    ->run('/path/to/audio-file');

Side note: Google only supports sending inline files that are: less than 10MB in size and less than 1 minute in length

Transcripts

By default the SpeechToText class will be passed the a BasicTranscript class. This class tells the SpeechToText class how to handle the response from the SpeechClient class from the Google Speech to Text Package. If you want to handle the response from the SpeechClient in in different way, e.g. including the word time offsets, you will need to set an other Transcript class that implements the TranscriptInterface.

// Using different types of transcripts (e.g. include word time offsets (startTime and endTime))
resolve(SpeechToText::class)->setTranscript(new WordTimeOffsets())
    ->run('gs://your-bucket-name/path-to-object');

Example output of WordTimeOffsets transcript

array:2 [
  'transcript' => array:10 [
      0 => array:3 [
        "transcript" => "hello world"
        "confidence" => 0.96761703491211
        "words" => array:9 [
          0 => array:3 [
            "word" => "hello"
            "startTime" => 0
            "endTime" => 0.3
          ]
          1 => array:3 [
            "word" => "world"
            "startTime" => 0.3
            "endTime" => 0.5
          ]
          ...
        ]
      ]
      1 => array:3 [
        "transcript" => "foo bar buz"
        "confidence" => 0.74065810441971
        "words" => array:7 [
            ...
        ]
      ]
  ]
  'words' => array:45 [
      0 => array:3 [
         "word" => "hello"
         "startTime" => 0
         "endTime" => 0.3
      ]
      1 => array:3 [
          "word" => "world"
          "startTime" => 0.3
          "endTime" => 0.5
      ]
      ...
  ]
]

Changelog

Please see CHANGELOG for more information what has changed recently.

Contributing

Contributions are welcome and will be fully credited. We accept contributions via Pull Requests on Github.

Pull Requests

PSR-2 Coding Standard - The easiest way to apply the conventions is to install PHP Code Sniffer.
Document any change in behaviour - Make sure the README.md and any other relevant documentation are kept up-to-date.
Create feature branches - Don't ask us to pull from your master branch.
One pull request per feature - If you want to do more than one thing, send multiple pull requests.

License

The MIT License (MIT). Please see License File for more information.

noardcode / speech-to-text

Maintainers

Details