Processing long audio

[v-nhandt21]

Which is the better way in processing long audio, for example, I have 60s audio, should I:

• Push whole audio for vocal-remover model to process it in batch (this takes much memory)

or

• Split the audio into 6 small pieces then run multiprocess put it into vocal remover

[Amirreza-Zeraati]

Hi there, The choice between processing a long audio file as a whole or splitting it into smaller pieces and processing them in parallel depends on various factors, including your available hardware resources, the specific vocal remover model and ...

• If you have ample memory and processing power, processing the whole audio at once may be more straightforward (In some cases, when the batch size is larger, the model learns better) However, if memory is limited, splitting and processing in parallel may be necessary.

• Check the vocal remover model's documentation and capabilities. Some models may be optimized for longer audio, while others may be designed for shorter segments. Choose a model that suits your chosen approach.

• Consider whether you need real-time processing. If you do, splitting and processing in parallel might be a better choice to reduce processing time.