Error:Maximum decoding length >0 while transcribing

[pranavbhat12]

For certain audios, I am getting this issue of Maximum decoding length >0 while generating the segments.

Parameters that were used: segments,info = whisper_model.transcribe( audio="sample.wav", task="translate", repetition_penalty=1.2, initial_prompt="Hello Yes Sir.Ignore silence or blank periods with focus on accurate transcription and punctuations.", # condition_on_previous_text=False, hotwords=["Set of Domain specific terms"], )

Error it shows in: generate_with_fallback-->self.model.generate(...)

Total prompt length with initial prompt and hotwords is 401 for this audio.

It is working for other audios for certain audios it is not able to generate.Tried to understand the code but could not figure it out.Even though max_length is 448 ideally it should say it is exceeding the max length of the model(this is not the case here though) but instead it says it must be greater than 0.How can the length go in negative?.Any help would be appreciated from the team.Thankyou. @MahmoudAshraf97 @Jiltseb