Hi.
Can I improve something here?
On Fri, Mar 30, 2018 at 2:43 AM, Rostislav Pehlivanov
wrote:
> On 30 March 2018 at 00:12, Даниил Чередник
> wrote:
>
> > Actually this test does not make a sense since encoder now use floating
> > point math. I should have noticed it bef
Actually this test does not make a sense since encoder now use floating
point math. I should have noticed it before.
Both patches have been updated. Checked on arm.
On Thu, Mar 29, 2018 at 1:52 AM, Michael Niedermayer wrote:
> On Tue, Mar 27, 2018 at 02:26:45AM +0300, Даниил Чередник wr
Now it should be fine. Second patch has been fixed.
On Mon, Mar 26, 2018 at 3:59 PM, Даниил Чередник
wrote:
> Hi.
> Yes, I reproduced it on mac. A bit strange, I run fate with valgrind
> during patch preparation, it was ok.
> I will fix it, and send updated patch.
>
> Thank yo
Hi.
Yes, I reproduced it on mac. A bit strange, I run fate with valgrind during
patch preparation, it was ok.
I will fix it, and send updated patch.
Thank you!
On Sun, Mar 25, 2018 at 5:57 PM, Michael Niedermayer wrote:
> On Sun, Mar 25, 2018 at 01:55:42PM +0300, Даниил Чередник wr
There are several reasons to replace dca psychoacoustic to common model:
- dca psychoacoustic has some quality problems especially at high
frequency bands
- unclear implementation
- aac implementation allows to use tonality measurement for future
improvements
- a bit faster
AAC psychoacoustic
New path attached.
Thanks.
On Sat, Jan 13, 2018 at 2:37 AM, James Almer wrote:
> On 1/12/2018 8:12 PM, Даниил Чередник wrote:
> > Hysterically dcaenc uses own implementation of time->frequency
> > transformation used by psychoacoustic. But actually function named fft in
&
Hysterically dcaenc uses own implementation of time->frequency
transformation used by psychoacoustic. But actually function named fft in
original dcaenc code is not fft. Power spectrum looks similar to mdct, and
Alexander E. Patrakov told me it is MDCT. But for me it is still a bit
strange, because
Hi.
Current opusenc_psy code (step_collect_psy_metrics) uses only last coeff in
range to calculate tone value. It looks like typo.
--
Daniil Cherednik
0001-opusenc_psy-Typo-use-all-coeffs-in-range-for-band-to.patch
Description: Binary data
___
ffmpeg
Hi.
This patch removes assert which can cause abort process in case of limited
bitrate:
Old behaviour - abort if at least one subband has 1bit quantizer
and consumed_bits still greater than frame_bits size. It was
a bit strange - we still could reduce bits consumption by reducing
SNR for other su
.
On Mon, May 8, 2017 at 7:59 AM, Rostislav Pehlivanov
wrote:
> On 6 May 2017 at 11:47, Даниил Чередник wrote:
>
> > Tiny fix: missed inline attr.
> >
> > On Thu, May 4, 2017 at 1:17 AM, Даниил Чередник >
> > wrote:
> >
> > > All notices were fixe
Tiny fix: missed inline attr.
On Thu, May 4, 2017 at 1:17 AM, Даниил Чередник
wrote:
> All notices were fixed. Also I found issue with uninitialized subband
> buffer - fixed. New patch attached.
>
> On Wed, May 3, 2017 at 7:49 AM, Rostislav Pehlivanov
> wrote:
>
>>
All notices were fixed. Also I found issue with uninitialized subband
buffer - fixed. New patch attached.
On Wed, May 3, 2017 at 7:49 AM, Rostislav Pehlivanov
wrote:
> On 2 May 2017 at 22:53, Даниил Чередник wrote:
>
> > Hi.
> >
> > This patch introduces initial impleme
Hi.
This patch introduces initial implementation of subband ADPCM encoding for
DCA codec.
Some results:
sample:
https://yadi.sk/d/B_3sVskM3HZiWK - original
https://yadi.sk/d/7CK47Nt63HZiWf - without adpcm
https://yadi.sk/d/25q1JDV93HZiWq - with adpcm
chirp tone:
https://yadi.sk/i/tZKHoJ1d3H
Hi.
My previous patch introduced Huffman for quantized data, this patch allows
to use Huffman for bit allocation indexes (abits). This is a less
significant improvement but still noticeable in case of limited bitrate. In
case of DIN_45403 input sample I got:
before:
Best PSNR is 44.47 for shift 0
On Sun, Jan 15, 2017 at 6:58 PM, Paul B Mahol wrote:
> On 1/15/17, Daniil CHerednik wrote:
> > Hi.
> > Is there any issue in this patch?
>
> Do you plan to work more on this encoder?
>
Yes I do.
>
> Please do not top post.
>
Ok, sorry.
>
> >
> > On Sun, Jan 8, 2017 at 1:22 AM, Daniil CHerednik
Hi.
Is there any issue in this patch?
On Sun, Jan 8, 2017 at 1:22 AM, Даниил Чередник
wrote:
> With real music and 256k bitrate encoding (the source was 44100, 16bit
> stereo) I got:
> Without Huffman: Best PSNR is 31.77 for shift 0
> With: Best PSNR is 37.45 for shift 0
.
I have attached new split set of patches.
On Sat, Jan 7, 2017 at 11:50 PM, Carl Eugen Hoyos
wrote:
> 2017-01-07 20:39 GMT+01:00 Rostislav Pehlivanov :
> > On 7 January 2017 at 16:11, Carl Eugen Hoyos wrote:
> >
> >> 2017-01-07 16:00 GMT+01:00 Даниил Чередник :
> >&
Hi!.
Currently I am working on improvement quality of DTS encoder. Following
patches introduce Huffman coding.
First one - reverse data layout
[SUBBAND_SAMPLES][DCAENC_SUBBANDS][MAX_CHANNELS] to
[MAX_CHANNELS][DCAENC_SUBBANDS][SUBBAND_SAMPLES]. I need it to write
more readable code, and reduce (a
with newly encoded ATRAC1 file. I included source pcm and
encoded aea in to the fate test. These files are here
https://yadi.sk/d/DA-xxzM_kcJ5t
On Wed, Nov 18, 2015 at 10:19 PM, Michael Niedermayer
wrote:
> On Tue, Nov 17, 2015 at 04:08:29AM +0300, Даниил Чередник wrote:
> > Than
de without
patch. I can share decoded wav file if needed.
On Mon, Nov 16, 2015 at 3:45 AM, Michael Niedermayer wrote:
> On Sat, Nov 14, 2015 at 02:59:30AM +0300, Даниил Чередник wrote:
> > Hello!
> >
> > I have noticed if decode
> >
> https://samples.ffmpeg.or
Hello!
I have noticed if decode
https://samples.ffmpeg.org/A-codecs/ATRAC1/Test%20tones%20disc%20-%20Chirp.aea
file by ffmpeg we got aliasing near 11025Hz. Screenshots:
https://yadi.sk/i/r-95jZkKkSnbu https://yadi.sk/i/PV92LNESkSnby
I was tried to solve it:
ATRAC1 - hybrid codec, has two stacked
21 matches
Mail list logo
