Google Cloud Platform (GCP) - speech api

This is an example using Google's speech to text api. A little more details can be found on the blog below.

http://jybaek.tistory.com/671

GCP Prerequisite

• Please install gcloud sdk first.
• Perform authentication and resolve associated dependencies.

Authentication

$ gcloud auth application-default login

Install Dependencies

$ pip install -r requirements.txt

Usage

Audio file recognition

To convert a file to text in its entirety, proceed as follows.

$ python speech.py
Transcript: 안녕 하세요 좋은 아침입니다

The default is to specify test.raw via the audio-path option. Take a look at the options through help as below.

$ python speech.py --help
usage: speech.py [-h] [--audio-path AUDIO_PATH]
                 [--language-code LANGUAGE_CODE]

speech to text

optional arguments:
  -h, --help            show this help message and exit
  --audio-path AUDIO_PATH
                        Audio file to convert to text.
  --language-code LANGUAGE_CODE
                        Language code. ( ko-KR, en-US, etc.. )

Here is an example of converting a file to streaming. The options are the same as for speech.py.

$ python speech_streaming.py
====================
transcript: 안녕 하세요 좋은 아침입니다
confidence: 0.5344622135162354

Real-time speech recognition

You need to install pypaudio, please refer to the link below to install it first.

https://stackoverflow.com/a/33821084/4599185

After the installation is completed, you can do the following. Speech recognition is pending, so deliver voice over the microphone.

$ python transcribe_streaming_mic.py

Most of the sample code that is registered with googlecloudplatform is used.