speechtotext.model.whisperWrapper

Modelwrapper implemented for whisper. Local and API.

API OPENAI_API_KEY and OPENAI_ORGANIZATION need to be in the ‘.env’.

Use this module like this:

# Imports
from speechtotext.model.whisperWrapper import *
from speechtotext.benchmark.benchmarks import *
from speechtotext.datasets import Dataset

# Create dataset
number_of_samples = 10
dataset = Dataset(path_to_dir="path/to/dir", name= "dataset_name")
id = "existing_audio_id"
number_of_samples = 10

# Create wrapper
whisperWrapper = WhisperWrapper(WhisperVersion.TINY)
# Add model to Plotting
# Get model
whisperWrapper.get_model()

# Benchmark choisen sample
whisperWrapper.benchmark_sample(dataset, id)

# Benchmark n random samples
array = whisperWrapper.benchmark_n_samples(dataset, number_of_samples)

Classes

WhisperAPIVersion	Enum for the available Whisper API models.
WhisperAPIWrapper	Wrapper for whisper API.
WhisperVersion	Enum for the available Whisper models.
WhisperWrapper	Wrapper for whisper model.