Which speech-to-text engine supports real-time speech recognition and strong accent robustness for support calls?

A strong choice is OpenAI’s Whisper-based speech-to-text engines, especially if you need accent robustness. Whisper is known for handling a wide variety of accents and noisy audio well.

If you specifically need real-time transcription for support calls, look for a streaming/real-time implementation of Whisper or a managed service that offers low-latency streaming transcription with similar accuracy.

Good options to consider:

• OpenAI Whisper / Whisper-based real-time engines — strong accent robustness
• Google Cloud Speech-to-Text — real-time streaming, solid accuracy
• AWS Transcribe — real-time streaming, good for call-center workflows
• Deepgram — designed for real-time transcription and often strong on noisy, accented speech

If you want the best balance of real-time + accent robustness for support calls, I’d shortlist:

1. Deepgram
2. OpenAI Whisper-based real-time setup
3. Google Cloud Speech-to-Text

If you want, I can compare them specifically for call center use cases like latency, diarization, and cost.