cockatoo-chain 0.1.0

The package is intended for STT, LLM and TSS models chaining.

cockatoo_chain

This repo is intended for STT, LLM and TSS models chaining.

Development Mode

Development Mode is intended for users who wish to contribute to the repo and thus needs to install additional dev-related packages for, e.g., code quality checking, to satisfy the standard of the repo. If you hope to contribute to the repository, please follow up the follow instructions:

Create virtual environment by poetry

# Force poetry to build virtual environment in the repo.
$ poetry config virtualenvs.in-project true

# Create virtual environment
$ poetry env use python

# Confirm the created virtual environment
$ poetry env info

Enter virtual environment and installed packages

# Get the command to enter virtual environment
$ poetry env activate
$ source activate <your venv>
$ make init-repo-setup

Model A Usage

From cockagoo_chain package, you can easily access the power of Model A (Speech-to-text) with a few lines of codes. We will learn how to from this section. For the time being, cockagoo_chain support below types of model A:

Name	Type	Supported language	Supported file type	Note
open_ai_whisper	Remote API	en, cn and more	File uploads are currently limited to 25 MB, and the following input file types are supported: mp3, mp4, mpeg, mpga, m4a, wav, and webm	Official doc

Get supported Model A options

You can use below code snippet to get the supported model A options:

>>> from cockatoo_chain.utils import model_a
>>> model_type = model_a.ModelType
>>> list(model_type)
[<ModelType.OPEN_AI_WHISPER: 'open_ai_whisper'>]

Transform input audio file into text

Below code snippet demonstrates how to obtain the OpenAI whipser wrapper for model A and apply it to transform the audio file into text:

>>> test_cn_audio_file_path = '~/test_audio_files/en_20240108_johnlee.wav'
>>> model_a_wrapper = model_a.get(model_type.OPEN_AI_WHISPER)
>>> response = model_a_wrapper.audio_2_text(test_cn_audio_file_path)
>>> response
Audio2TextData(
    text='Hello, this is for testing in English. We will use this to evaluate model SST...',
    spent_time_sec=7.87896990776062,
    audio_file_path='/root/test_audio_files/en_20240108_johnlee.wav')
>>> response.text
'Hello, this is for testing in English. We will use this to evaluate model SST and see how it performs. Thanks.'