Brunopaes / gemini-replicator

This project serves as a transcriber for short - seconds length - and long audios - up 10 minutes audios and down 30 minutes audio.

Gemini Replicator: Speech-to-text decoder

Optimized for python 3.6

This project serves as a transcriber for short - seconds length - and long audios - up 10 minutes audios and down 30 minutes audio. It uses some services from Google Cloud.

---

Dependencies

For installing the requirements, in your venv or anaconda env, just run the following command:

pip install -r requirements.txt

---

Project's Structure

.
└── evil-eye
    ├── data
    │   ├── sample_audio_1.mp3
    │   ├── ...
    │   └── sample_audio_2.mp3
    ├── docs
    │   └── CREDITS
    ├── src
    │   ├── __init__.py
    │   ├── settings.json
    │   └── speech.py
    ├── tests
    │   └── unittests
    │       └── __init__.py
    ├── .gitignore
    ├── LICENSE
    ├── README.md
    └── requirements.txt

Directory description

• data: The data dir. Group of non-script support files.
• docs: The documentation dir.
• src: The scripts & source code dir.
• tests: The unittests dir.

---

Usage Notes

Section aimed on clarifying some running issues.

Running

For running it, at the ~/src directory just run:

python speech.py

or, if importing it as a module, just run:

from speech import local_short_recognition, cloud_long_recognize

if __name__ == '__main__':
    local_short_recognition('path/to/file')
    cloud_long_recognize('gs://[bucket]/[file]')

JSON structure

{
  "type": "",
  "project_id": "",
  "private_key_id": "",
  "private_key": "",
  "client_email": "",
  "client_id": "",
  "auth_uri": "",
  "token_uri": "",
  "auth_provider_x509_cert_url": "",
  "client_x509_cert_url": ""
}

obs: you must create your own Google's service account key - just follow this instructions obs: in order to run this application you must have a json file at ~/src/settings.json. This json must follow the structure above.