Skip to content

Wrong update of prev_Q_syn

Basic info

  • Fixed point:

Bug description

While working on #1232, we found that the update of prev_q_syn is wrongly done which caused Q_syn to almost always be -1 leading to a very poor ACELP synthesis quality for low level signals as show below:

image

Ways to reproduce

see information of #1232 and check the first transport layer 8.4 and 8.5s to have an idea.