2013-09-06 02:21 AM
I am using the sample application Audio playback and recording available along with the STM32F4 evaluation kit. The sample application will save the recorded file to USB memory stick. Currently we are using this recorded file for speech processing. We are observing that the recorded file have noise and lot of glitches between the words. Also we can observe missing words in the speech.
For eg: if we are speaking the sentence '' We are testing 1 2 3 4 5'' the recorded file play like ''we are te1 2 3 4 5''. Can you give some suggestion on how to improve this sound quality by removing missing words and glitches.2013-09-06 02:47 AM
Your USB mass storage device probably buffers a certain amount of blocks, until it actually writes the data. That can take several hundreds of milliseconds, were the device remains busy erasing/writing the flash.
Check that your application actually pays attention to the device status. And design you application in a way to cope with this delays - there are a lot of tutorials about buffering strategies around for those cases.We are observing that the recorded file have noise ...That noise is most certainly introduced in the analogue section of your sampling hardware. I wouldn't care too much for a prototype. However, your evaluation hardware might be not the best platform for high-class audio applications.
2014-03-05 12:30 AM
I have my STMF4 Discovery board working as USB Audio Sound Card (not with USB stick, as native PC USB sound card).
Works great. Also UART mode (VCD) works for me. Project can be downloaded here: http://www.tjaekel.com Best regards Torsten