Unmatched Accuracy and Lightning Speed in Python for Speech Recognition

---

---

AI Nuggets

Assembly AI API Tutorial Outline

Getting Started with Assembly AI’s Universal One Model

1. Import Assembly AI’s Python SDK
   • Command: import assemblyai as aai
   • If not installed: pip install assemblyai
2. Set Assembly AI API Key
   • Create an account at assemblyai.com to get an API key.
3. Define Audio Data
   • Provide a URL to an audio or video file or upload a file from your system.
4. Transcribe Audio Data
   • Define a transcriber object.
   • Call the transcribe function with the audio URL.
   • To get only the transcription text: transcript.text

Customizing Output with Assembly AI API

1. Using Universal One Model
   • By default, the API uses the Universal One model when calling the transcribe function.
2. Using Nano Tier for Bulk Transcription
   • Pass a config argument to the transcriber with speech_model set to ‘nano’.
   • Example: transcriber.transcribe(audio_url, config={'speech_model': 'nano'})
3. Disabling Punctuation and Formatting
   • Set punctuate to false and format_text to false in the config.
   • Example: transcriber.transcribe(audio_url, config={'punctuate': False, 'format_text': False})
4. Transcribing Non-English Audio Files
   • Set language_code to the desired language code.
   • For automatic language detection, set automatic_language_detection to true.
   • Example: transcriber.transcribe(audio_url, config={'language_code': 'es', 'automatic_language_detection': True})
5. Getting Speaker Labels
   • Set speaker_labels to true in the config.
   • Print the results of the speaker diarization.
   • Example: transcriber.transcribe(audio_url, config={'speaker_labels': True})

Pricing and Usage Tracking

• Pricing
  • Best tier: $0.37 per hour.
  • Nano tier: $0.12 per hour.
• Usage Tracking
  • Visit the Assembly AI dashboard to track transcription hours and costs.

Additional Resources

• Assembly AI Documentation
  • URL: assemblyai.com/docs
• YouTube Tutorials
  • Description mentions links to tutorials in the video description.
• Next Steps
  • To learn how to transcribe a stream of audio with Assembly AI, watch the suggested video.

Notes

• The code examples and configurations can be found in the Assembly AI documentation under the speech recognition section.
• The tutorial demonstrates how to use the Assembly AI Python SDK to transcribe audio data and customize the transcription output.
• The tutorial also covers how to use different tiers of the Assembly AI service and how to handle audio data in different languages.
• The pricing information and how to track usage are provided for users to manage their costs effectively.