Difference in MDCT for LTP |
![]() ![]() |
Difference in MDCT for LTP |
Sep 3 2011, 06:21
Post
#1
|
|
|
Group: Members Posts: 56 Joined: 19-January 11 Member No.: 87451 |
hi all,
In case of AAC with LTP in decoder,after predicting the current frame by using the equation x_est(i) = ltp_coef*x_rec(i – M – ltp_lag) we get the N predicted time samples(N,length of window transform) After this we have to convert it to frequency domain using MDCT.. here can we use the same MDCT as in encoder for converting the time samples we read from the .wav file to f-domain, or is there any difference between the normal MDCT we use and the MDCT we use for LTP?? Please help Regards Pratheek |
|
|
|
Sep 9 2011, 10:45
Post
#2
|
|
|
Group: Members Posts: 56 Joined: 19-January 11 Member No.: 87451 |
somebody please reply
|
|
|
|
Sep 9 2011, 15:25
Post
#3
|
|
|
Group: Members Posts: 107 Joined: 3-April 09 Member No.: 68627 |
I'm not an expert, but as far as I understand the specification, it should be the same transform that is used for conventional long blocks encoding.
If you are interested in actual implementation, you could take a look at the FAAC source code. This post has been edited by alexeysp: Sep 9 2011, 15:28 |
|
|
|
Jan 22 2012, 00:40
Post
#4
|
|
|
Group: Members Posts: 1 Joined: 22-January 12 Member No.: 96606 |
hi all, In case of AAC with LTP in decoder,after predicting the current frame by using the equation x_est(i) = ltp_coef*x_rec(i – M – ltp_lag) we get the N predicted time samples(N,length of window transform) After this we have to convert it to frequency domain using MDCT.. here can we use the same MDCT as in encoder for converting the time samples we read from the .wav file to f-domain, or is there any difference between the normal MDCT we use and the MDCT we use for LTP?? Please help Regards Pratheek In the decoder, don't you need to do IMDCT? Therefore you cannot used MDCT in the encoder. Or I may not understand your question clearly |
|
|
|
![]() ![]() |
|
Lo-Fi Version | Time is now: 20th May 2013 - 03:14 |