The faster-whisper-openai-api is an audio transcription service based on the Faster-Whisper model. It supports the transcription of audio files into text and offers multiple output formats, referencing the OpenAI api.
This project involves a web application that utilizes an API endpoint /v1/audio/transcriptions to transcribe audio files with different supported formats.
To make a request to this API, send a POST request to /v1/audio/transcriptions with the audio file you want to transcribe.
The API accepts the following parameters:
file: The audio file to be transcribed.model: The model to be used for transcription, default is 'whisper-1'.language: The language of the audio file.prompt: The initial prompt.response_format: The format you want the transcription to be returned in. Options include 'json', 'text', 'srt', 'verbose_json', 'vtt'.temperature: A float value for the temperature parameter.timestamp_granularities[]: A list of timestamp granularities.
The API will return a transcription of the audio in the requested format. If there is an error processing the request, the API will return a 400 or 500 error with an error message.