Edit

What is the Whisper model?

  • Article

The Whisper model is a speech to text model from OpenAI that you can use to transcribe audio files. The model is trained on a large dataset of English audio and text. The model is optimized for transcribing audio files that contain speech in English. The model can also be used to transcribe audio files that contain speech in other languages. The output of the model is English text.

Whisper models are available via the Azure OpenAI Service or via Azure AI Speech. The features differ for those offerings. In Azure AI Speech, Whisper is just one of several speech to text models that you can use.

You might ask:

  • Is the Whisper Model a good choice for my scenario, or is an Azure AI Speech model better? What are the API comparisons between the two types of models?

  • If I want to use the Whisper Model, should I use it via the Azure OpenAI Service or via Azure AI Speech? What are the scenarios that guide me to use one or the other?

Whisper model or Azure AI Speech models

Either the Whisper model or the Azure AI Speech models are appropriate depending on your scenarios. If you decide to use Azure AI Speech, you can choose from several models, including the Whisper model. The following table compares options with recommendations about where to start.

Scenario Whisper model Azure AI Speech models
Real-time transcriptions, captions, and subtitles for audio and video. Not available Recommended
Transcriptions, captions, and subtitles for prerecorded audio and video. The Whisper model via Azure OpenAI is recommended for fast processing of individual audio files. The Whisper model via Azure AI Speech is recommended for batch processing of large files. For more information, see Whisper model via Azure AI Speech or via Azure OpenAI Service? Recommended for batch processing of large files, diarization, and word level timestamps.
Transcript of phone call recordings and analytics such as call summary, sentiment, key topics, and custom insights. Available Recommended
Real-time transcription and analytics to assist call center agents with customer questions. Not available Recommended
Transcript of meeting recordings and analytics such as meeting summary, meeting chapters, and action item extraction. Available Recommended
Real-time text entry and document generation through voice dictation. Not available Recommended
Contact center voice agent: Call routing and interactive voice response for call centers.​ Available Recommended
Voice assistant: Application specific voice assistant for a set-top box, mobile app, in-car, and other scenarios. Available Recommended
Pronunciation assessment: Assess the pronunciation of a speaker's voice. Not available Recommended
Translate live audio from one language to another. Not available Recommended via the speech translation API
Translate prerecorded audio from other languages into English. Recommended Available via the speech translation API
Translate prerecorded audio into languages other than English. Not available Recommended via the speech translation API

Whisper model via Azure AI Speech or via Azure OpenAI Service?

If you decide to use the Whisper model, you have two options. You can choose whether to use the Whisper Model via Azure OpenAI or via Azure AI Speech. In either case, the readability of the transcribed text is the same. You can input mixed language audio and the output is in English.

Whisper Model via Azure OpenAI Service might be best for:

  • Quickly transcribing audio files one at a time
  • Translate audio from other languages into English
  • Provide a prompt to the model to guide the output
  • Supported file formats: mp3, mp4, mpweg, mpga, m4a, wav, and webm

Whisper Model via Azure AI Speech might be best for:

  • Transcribing files larger than 25MB (up to 1GB). The file size limit for the Azure OpenAI Whisper model is 25 MB.
  • Transcribing large batches of audio files
  • Diarization to distinguish between the different speakers participating in the conversation. The Speech service provides information about which speaker was speaking a particular part of transcribed speech. The Whisper model via Azure OpenAI doesn't support diarization.
  • Word-level timestamps
  • Supported file formats: mp3, wav, and ogg
  • Customization of the Whisper base model to improve accuracy for your scenario (coming soon)

Regional support is another consideration.

  • The Whisper model via Azure OpenAI Service is available in the following regions: EastUS 2, India South, North Central, Norway East, Sweden Central, and West Europe.
  • The Whisper model via Azure AI Speech is available in the following regions: Australia East, East US, North Central US, South Central US, Southeast Asia, UK South, and West Europe.

Next steps