I have posted about this problem before. Evidently it was fixed and then broke again.
Digital audio recorders’ and camcorders’ speed is stable enough to run more than one at the same time unsynchronized with each other to create stereo or more often usefully, surround sound, however, the reccorders’ speed will differ slightly (or they may even have different nominal sample rates). It is necessary to apply the pitch and tempo effect so the recordings synchronize with the needed accuracy. The effect, however, does not return a result with the exact number of samples requested. My test recording, which you can duplicate, is a default tone 100,000,000 samples long at 48 K – 2083.333 seconds. When I ask Pitch and Tempo to expand it to 100,050,000 samples, I get a result of 100,050 025 samples. 25 Samples at 48K is 0.52 millisecond, which is enough to skew a stereo image. When I ask for 100,100,000 samples, I get 100,100,100 samples. When I ask for 100,200,000, I get 100,200,401 and when I ask for 100,400,000, I get 100,399,590. Apparently there is some rounding of numbers or conversion from samples to other less precise units before the processing. Shortening rather than lengthening, when I ask for 99,950,000 samples, I get 99,950,025. You can try with other values. My previous report is linked below. I am running Audacity 3.7.7 under Windows 10 on a machine with an AMD 3900X CPU.