Raw recorded file
AI Models & Assistants / Knowledge library
Whisper
General-purpose speech recognition model
Why learn Whisper?
Offers the documented OpenAI model capabilities through hosted.
The problem it solves
Provides model inference for supported audio inputs.
Where it came from
Whisper: General-purpose speech recognition model Published aliases: whisper-1. Input: audio. Output: text. Checked 2026-10-13. Model capabilities are distinct from tools available in a particular agent.
Build your understanding
Work through these exercises in order. Try the task before opening a reference.
- Explain what Whisper does.
- Identify a use case for Whisper.
- Describe a limitation of Whisper.
- Name a compatible tool for Whisper.
- Compare Whisper against an alternative.
Documentation and sources
- Primary source / model card (developers.openai.com)
Catalog research notes
- See the AI Workbench for source dates, pricing evidence and configuration details.
Connect the ideas
Built on
Real World Example
Automated Legal Deposition Indexing
A regional law firm managing high volumes of recorded legal depositions.
The business problem
Manual transcription of hours-long audio files is labor-intensive, prone to human error, and creates significant delays in case preparation.
How this topic solves it
The firm utilizes the Whisper-1 model to programmatically convert audio recordings into structured text, enabling rapid search and keyword identification across deposition archives.
- Capture raw audio files from deposition sessions in supported formats.
- Submit audio files to the whisper-1 transcription endpoint via API.
- Receive the generated text output for indexing in the firm's internal database.
- Perform keyword analysis on the transcribed text to identify key testimony.
The flow of audio data from capture to searchable text.
Transcription engine
Transcribed testimony
Searchable archive
- Deposition Audio → submits audio → Whisper-1 API
- Whisper-1 API → returns text → Text Output
- Text Output → stores for search → Legal Database
Implementation example
This code demonstrates the basic API call to the transcription endpoint, which maps directly to the solution of converting audio files into searchable text.
import openai
# Send audio file to whisper-1 for transcription
audio_file = open("deposition_001.mp3", "rb")
transcript = openai.Audio.transcribe("whisper-1", audio_file) # Calls the transcription endpointExpected business outcome
Legal teams gain immediate access to searchable text records, significantly reducing the time required to prepare for trial proceedings.
Limitations and checks
Whisper-1 is a general-purpose model and may struggle with highly specialized legal jargon or thick regional accents, which can be verified by conducting spot-checks against the original audio.