How to beginner · 3 min read

How to install Whisper locally

Q: How to install Whisper locally

To install Whisper locally, use pip install openai-whisper to get the official package. Then load the model with whisper.load_model() and transcribe audio files directly on your machine without API calls.

Quick answer

To install Whisper locally, use pip install openai-whisper to get the official package. Then load the model with whisper.load_model() and transcribe audio files directly on your machine without API calls.

PREREQUISITES

Python 3.8+
pip install openai-whisper
ffmpeg installed and in system PATH

Setup

Install the openai-whisper package via pip and ensure ffmpeg is installed on your system for audio processing.

bash

pip install openai-whisper

Step by step

Use the following Python code to load the Whisper model and transcribe an audio file locally.

python

import whisper

# Load the base Whisper model
model = whisper.load_model("base")

# Transcribe an audio file
result = model.transcribe("audio.mp3")

print(result["text"])

output

This is the transcribed text from the audio file.

Common variations

Use different model sizes like tiny, small, medium, or large by changing load_model("model_name").
For faster transcription on CPU, use smaller models.
Use faster-whisper for optimized performance on CPUs.

python

import whisper

# Load a smaller model for faster CPU transcription
model = whisper.load_model("small")

result = model.transcribe("audio.mp3")
print(result["text"])

output

Transcribed text from audio using the small model.

Troubleshooting

If you get an error about ffmpeg, install it and ensure it's in your system PATH.
For Windows, download ffmpeg from the official site and add to PATH.
If transcription is slow, try a smaller model or run on GPU if available.

✅

Key Takeaways

Install Whisper locally with pip install openai-whisper and ffmpeg.
Load models like base or small for transcription without API calls.
Use smaller models for faster CPU performance or GPU for best speed.
Ensure ffmpeg is properly installed to avoid runtime errors.

Verified 2026-04 · whisper-1, openai-whisper

Verify ↗