Yes, frustratingly we don't have good speech-to-text (STT/ASR) to transcribe such differences.
I recently finetuned a TTS* to be able to emit laughter and hunting for transcriptions which include non-verbal sounds was the hardest part of it. Whisper and other popular transcription systems will ignore sigh, sniff, laugh, etc and can't detect mispronounciations etc.
