speechtotext.benchmark.customBenchmarks

Module for benchmarks of speechtotext.

Use this module like this:

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

# Settings
number_of_samples = 10
dataset = Dataset(path_to_dir="path/to/dir", name= "dataset_name")
Benchmark.set_dataset(dataset)

# Create benchmark
wb = WhisperBenchmark()

# Run benchmark
wb(number_of_samples)

# Convert metrics to pandas dataframe
df = wb.convert_to_pandas()
print(df)

# Save metrics to csv (saves with datetime in name)
benchmark_results_to_csv([wb])

Classes

AmazonAPIBenchmark	Benchmark for Amazon API transcribe.
AssemblyAIAPIBenchmark	Benchmark for AssemblyAI API.
AzureAPIBenchmark	Benchmark for Azure API.
DeepgramAPIBenchmark	Benchmark for Deepgram API.
GoogleAPIBenchmark	Benchmark for Google API transcribe.
SpeechmaticsAPIBenchmark	Benchmark for Speechmatics API.
WhisperAPIBenchmark	Benchmark for API whisper models.
WhisperBenchmark	Benchmark for local whisper models.