At present, the core uploads samples in batches of ~64, which means the audio batch callback is used many (~9) times per frame. This 'overstresses' the frontend audio buffer and leads to bad AV synchronisation.
This PR fixes the issue by ensuring that the audio batch callback is used to send all available samples only once per frame.
At present, the core uploads samples in batches of ~64, which means the audio batch callback is used many (~9) times per frame. This 'overstresses' the frontend audio buffer and leads to bad AV synchronisation.
This PR fixes the issue by ensuring that the audio batch callback is used to send all available samples only once per frame.