Whisper API — Speech to Text with OpenAI

Introduction

Whisper converts speech to text with high accuracy. This guide covers API integration.

• Getting Started
• Features
• Supported Formats
• Pricing
• Tips
• Use Cases
• Limitations
• Conclusion
• FAQ

Getting Started

from openai import OpenAI

client = OpenAI(api_key="sk-...")

with open("audio.mp3", "rb") as f:
    transcript = client.audio.transcriptions.create(
        model="whisper-1",
        file=f
    )

print(transcript.text)

Features

• Multiple languages
• High accuracy
• Automatic punctuation
• Timestamps available
• Reasonable pricing

Supported Formats

MP3, MP4, MPEG, MPGA, M4A, WAV, WEBM

Pricing

$0.006 per minute, very affordable.

Tips

• Clean audio produces better results
• Specify language for improvement
• Handle errors gracefully
• Cache transcripts
• Monitor API usage

Use Cases

Transcription services, voice commands, accessibility, meeting notes, customer feedback.

Limitations

• Audio length limit (25MB)
• No real-time streaming
• Language detection sometimes inaccurate

Conclusion

Whisper excellent choice for speech-to-text needs.

FAQ

Q: Accuracy? A: Very high, comparable to human transcription.

Q: Other languages? A: Supports 99+ languages with good accuracy.