Opened 4 years ago

Closed 2 years ago

Last modified 22 months ago

#9270 closed defect (fixed)

8-Bit FLAC Playback Skips Silent Portions

Reported by: nikku4211 Owned by:
Priority: important Component: avcodec
Version: git-master Keywords: FLAC regression
Cc: nikku4211 Blocked By:
Blocking: Reproduced by developer: yes
Analyzed by developer: no

Description

Summary of the bug: When using FFPlay to play an 8-bit .FLAC, its silent portions are often skipped over during playback.

I use programs like MPC-HC that use FFMPEG as a basis for audio and video playback.
I have consistently noticed that these programs tend to play 8-bit .FLACs incorrectly, skipping over silent portions, which can mess up timing for music, audio skits, tests, etc.

How to reproduce:

% ffplay -i mptrendertest.flac

ffmpeg version 2021-05-26-git-7a879cce37
built on 2021-05-26

Attachments (2)

MPTRenderTest.flac (42.9 KB ) - added by nikku4211 4 years ago.
An 8-bit FLAC file showing a beat. On FFPlay and FFMPEG-based programs, the silent portions, crucial to timing, are skipped.
ffplay-20210529-211506.log (27.5 KB ) - added by nikku4211 4 years ago.
The .log output from FFPlay when playing the sample .FLAC file included.

Download all attachments as: .zip

Change History (8)

by nikku4211, 4 years ago

Attachment: MPTRenderTest.flac added

An 8-bit FLAC file showing a beat. On FFPlay and FFMPEG-based programs, the silent portions, crucial to timing, are skipped.

by nikku4211, 4 years ago

Attachment: ffplay-20210529-211506.log added

The .log output from FFPlay when playing the sample .FLAC file included.

comment:1 by Carl Eugen Hoyos, 4 years ago

Component: ffplayavcodec
Keywords: regression added; 8-bit ffplay removed
Priority: normalimportant
Reproduced by developer: set

For future tickets: Please do not report issues against ffplay if they are reproducible with ffmpeg.

Regression since 60a684939ab61dbfdf55105907808e015e5f6ec2

in reply to:  1 ; comment:2 by Balling, 4 years ago

Replying to Carl Eugen Hoyos:

Regression since 60a684939ab61dbfdf55105907808e015e5f6ec2

Is that the same regression that is in #9185?? That means our flac is not lossless at all. Wow.

in reply to:  2 ; comment:3 by Carl Eugen Hoyos, 4 years ago

Replying to Balling:

Replying to Carl Eugen Hoyos:

Regression since 60a684939ab61dbfdf55105907808e015e5f6ec2

Is that the same regression that is in #9185??

Are you asking if 60a684939ab61dbfdf55105907808e015e5f6ec2 and 9300de0409d52272387a9b9d84143dba212291f4 are the same commit?

in reply to:  3 comment:4 by Balling, 4 years ago

Replying to Carl Eugen Hoyos:

Are you asking if 60a684939ab61dbfdf55105907808e015e5f6ec2 and 9300de0409d52272387a9b9d84143dba212291f4 are the same commit?

So two different regressions, wow.

comment:5 by Balling, 3 years ago

maybe those silence portions have crc16 mismatch?? I thought we do not verify crc16.

ffmpeg.exe -i MPTRenderTest.flac -c:a pcm_s8 -f md5 -

prints md5 as in flac head now after 2ed5925e26f77fc377cedcefb3864dc1a4712210

Last edited 22 months ago by Balling (previous) (diff)

comment:6 by Elon Musk, 2 years ago

Resolution: fixed
Status: newclosed
Note: See TracTickets for help on using tickets.