When using time-stretching (WSOLA algorithm) the lenght of output have incorrect size, generating an audio with a length of a few milliseconds less, and other times with a few milliseconds more. For example. An audio that is 1000 milliseconds long, when time-stretched with tempo=0.5, should generate an audio that is 2000 milliseconds long, but actually generates an audio that has 2025 ms, sometimes 1980 ms. This varies depending on the parameters and the audio lenght. The correct size is rarely achieved. Any suggestion on how to fix?
When using time-stretching (WSOLA algorithm) the lenght of output have incorrect size, generating an audio with a length of a few milliseconds less, and other times with a few milliseconds more. For example. An audio that is 1000 milliseconds long, when time-stretched with tempo=0.5, should generate an audio that is 2000 milliseconds long, but actually generates an audio that has 2025 ms, sometimes 1980 ms. This varies depending on the parameters and the audio lenght. The correct size is rarely achieved. Any suggestion on how to fix?