After using the VAD model to segment the recording, I called the ProcessAsync method with samples from a small segment of the recording, but the recognition result was incorrect.
But I passed all samples through SelectMany (x=>x) After ToArray() is connected, it can be successfully processed.
But after running for a period of time, the hallucinations became severe.Repeatedly output the following content
After using the VAD model to segment the recording, I called the ProcessAsync method with samples from a small segment of the recording, but the recognition result was incorrect.
But I passed all samples through SelectMany (x=>x) After ToArray() is connected, it can be successfully processed.
But after running for a period of time, the hallucinations became severe.Repeatedly output the following content![image](https://github.com/sandrohanea/whisper.net/assets/12029452/c355759d-15b2-4ac9-8d53-401fb1e88048)