Hello, when we used lyra 1.3.2, we found that the codec was processed by down-sampling 16K internally, and then up-sampled to the target bit rate. At present, the internal 16k processing will cause the output spectrum to be truncated, which will greatly damage the music sound. Therefore, if we set kInternalSampleRateHz to 32k or 48k, a crash will occur in linux enviroment.
INFO: Created TensorFlow Lite XNNPACK delegate for CPU.
double free or corruption (!prev)
I would like to ask whether this crash problem can be solved. If it cannot be solved temporarily, will Lyra provide API to support 32k/48k codec processing?
Hello, when we used lyra 1.3.2, we found that the codec was processed by down-sampling 16K internally, and then up-sampled to the target bit rate. At present, the internal 16k processing will cause the output spectrum to be truncated, which will greatly damage the music sound. Therefore, if we set kInternalSampleRateHz to 32k or 48k, a crash will occur in linux enviroment.
INFO: Created TensorFlow Lite XNNPACK delegate for CPU. double free or corruption (!prev)
I would like to ask whether this crash problem can be solved. If it cannot be solved temporarily, will Lyra provide API to support 32k/48k codec processing?