lt-dictation-to-text

Data

data\sentences

Single file contains single audio sentence whithout repetitions. Used for further comparison analysis.

data\sentences_repeated

Here we have dictation, full sentence at the beginning and at the end + phrase repetitions. Not used for further comparison analysis. Transcription and confidence levels can be found in data\raw_detailed_output.md

An example of a single audio file:

• This is a sentence about my city
• This is
• a sentence
• about my city
• This is a sentence about my city

Setup

• GCP plan required
• gsutil required to move audio files to Google Storage
• python -m venv venv
• venv\Scripts\activate.bat
• pip install -r requirements.txt
• SET GOOGLE_APPLICATION_CREDENTIALS=[PATH_to_google_service_account_json]

TODO

1. Use sentences_repeated and phrases for better results