Google Speech API

This project use Google Cloud Speech-to-Text enables developers to convert audio to text by applying powerful neural network models in an easy to use API.

Before Installation

We need to install:

• Git
• Node
• Docker(CE)

Installation

Create a folder speech-api inside this folder clone these below two projects.

git clone [email protected]:jiue123/laradock_customized.git
git clone [email protected]:jiue123/google_speech_project.git

Append the below content to the end of host file for assigning domain to local ip

(For windows 10 the file is located in C:\Windows\System32\drivers\etc\hosts)

(For Mac the file is located in /etc/hosts)

127.0.0.1 audio.local

Step 1: Config laradock

Go to the folder laradock_customized.

Switch branch to feature/audio_convert.

Copy .env-example to .env file.

Inside folder nginx/sites/, copy laravel.conf.example file to audio.conf file.

Open audio.conf file and edit these below lines, then save.

server_name audio.local;
root /var/www/google_speech_project/public;

Inside folder laradock_customized run command:

docker-compose up -d nginx mysql php-worker phpmyadmin

Use command docker ps -a to check if container mysql and nginx is running or not. If running it's ok.

Step 2: Config google-speech project

Go to the folder google_speech_project.

Switch branch to develop.

Copy .env-example to .env file.

Edit these below lines of .env file.

APP_URL=http://audio.local

DB_CONNECTION=mysql
DB_HOST=mysql
DB_PORT=3306
DB_DATABASE=default
DB_USERNAME=root
DB_PASSWORD=root

#Step 2.1

Go to the folder laradock_customized.

Run command:

docker-compose exec workspace bash

Inside container workspace run command:

cd /var/www/google_speech_project

composer install

php artisan key:generate

php artisan migrate

php artisan db:seed

exit

#Step 2.2

Go to the folder google_speech_project.

Run command:

npm install

npm run dev

Step 3 Enable google services and create service account key

#Step 3.1 Enable services needed

Cloud Speech API

Google Cloud Storage

After enable Google Cloud Storage, go to https://console.cloud.google.com/storage and create a bucket

Example bucket name is bucket0891

#Step 3.2 Create service account key.

Follow step by step in below link for create service account key. Setting up authentication

After have json file, create folder .credentials inside folder google_speech_project and put the json file on this folder.

#Step 3.3 Config .env file for provide authentication credentials to your application

Open .env file inside folder google_speech_project

Find rows as below:

#Google Credentials
GOOGLE_PROJECT_ID=
GOOGLE_APPLICATION_CREDENTIALS=

Type google project ID

Example: GOOGLE_PROJECT_ID=speech-api-111111

Type path to json file inside the folder .credentials

Example: GOOGLE_APPLICATION_CREDENTIALS=".credentials/Speech_API-12ab34cd56.json"

#Step 3.4 Fill information bucket of google storage service

Find rows as below:

#Google Storage
USE_GOOGLE_STORAGE=
STORAGE_BUCKET_NAME=
STORAGE_BUCKET_URL=

Set value for USE_GOOGLE_STORAGE by true.

USE_GOOGLE_STORAGE=true

Type bucket name.

Example: STORAGE_BUCKET_NAME=bucket0891

Set value for STORAGE_BUCKET_URL

STORAGE_BUCKET_URL=http://storage.googleapis.com/

Test

Open browser and type http://audio.local to see the result.

user: [email protected]

pass: secret

Note

• This project have storage service s3(AWS) and Blob(Azure). You can config for use the services in file .env inside folder google_speech_project

   FILESYSTEM_DRIVER=
   
   #Microsoft Azure
   AZURE_BLOB_STORAGE_ENDPOINT=
   AZURE_BLOB_STORAGE_CONTAINER=
   AZURE_BLOB_SERVICE_URL=
   
   #AWS Amazon
   AWS_ACCESS_KEY_ID=
   AWS_SECRET_ACCESS_KEY=
   AWS_DEFAULT_REGION=
   AWS_VERSION=
   AWS_BUCKET=
   AWS_BUCKET_URL=

#With s3 service

FILESYSTEM_DRIVER=s3

#With Blob service

FILESYSTEM_DRIVER=azure

If use s3(AWS) or Blob(Azure) must set USE_GOOGLE_STORAGE by false or blank.

• But i recommend using the google storage service because with longer audio more than 1 minutes you must use Google Cloud Storage objects as input.
• In this project just support convert speech to text with audio file with encoding .wav(8000 Hz) and .flac(16000 Hz). All encodings support only 1 channel (mono) audio.

You can find out more encoding for audio file in link AudioEncoding

Add the encoding you want into google_speech_project/config/google.php.

google_speech_options_convert' => [
    'wav' => [
        'encoding' => 'MULAW',
        'sampleRateHertz' => 8000,
    ],
    'flac' =>[
        'encoding' => 'FLAC',
        'sampleRateHertz' => 16000,
    ],
]