Skip to content

openai_whisper_transcription-README.md

Latest commit

 

History

History
48 lines (31 loc) · 1.85 KB

openai_whisper_transcription-README.md

File metadata and controls

48 lines (31 loc) · 1.85 KB
author OpenAI GPT-4
date 2023-05-19
filename README.md

OpenAI Whisper Transcript Generator

This is a Bash script that uses the OpenAI Whisper API to transcribe audio files. It supports various input file formats, includes file chunking for larger inputs, and provides detailed error logging.

Requirements

  • ffmpeg
  • curl
  • jq (for parsing JSON responses)

You can install these with the following command (for Arch Linux):

sudo pacman -S ffmpeg curl jq

Features

  • Environment variable for OpenAI token (OPENAI_API_KEY). If not defined, the script will notify the user.
  • Support for different output modes: file output (-o outputfile) and standard output (-stdout or -so). If both are specified, the script will save to a file and output to stdout.
  • Help message provided if no parameters are supplied or if "-h" or "--help" is used.
  • Detection of input file format and conversion to a format accepted by Whisper (if necessary).
  • Logging of billing information into a specific log file. The log file path can be specified by the environment variable OPENAI_WHISPER_BILLING_LOG. If not defined, the default log file will be "${HOME}/.openai_whisper_billing.log".
  • Optional flag -p/--prompt to specify a prompt to improve transcription quality.
  • Optional flag --promptfile to specify a text file containing the prompt.
  • Optional flag -l/--lang/--language to specify the language of the input audio.
  • If interrupted with Ctrl-C, captures signal with a trap and removes temporary files.

Usage

bashCopy code

./whisper_transcript.sh -o outputfile inputfile

For detailed usage instructions, use the help command:

bashCopy code

./whisper_transcript.sh -h

Note

For input files larger than 25MB, the script will automatically split the file into chunks of 20MB each (at most) and generate separate output files for each chunk.