VideoHelp Forum
+ Reply to Thread
Page 2 of 5
FirstFirst 1 2 3 4 ... LastLast
Results 31 to 60 of 122
Thread
  1. Originally Posted by rowjekto View Post
    Originally Posted by hokkom View Post
    What I would like to know is what is the best settings or a good starting place for doing this for audacity.
    Here is a suggestion I've found for Audacity's Compressor settings:

    - Threshold: -30.0 dB
    - Noise Floor: -50.0 dB
    - Ratio: 4:1
    - Attack Time: 0.3 sec.
    - Release Time: 3.0 sec.
    - "Make-up gain for 0dB after compressing": enabled

    The result is OK, but the loud parts are still a bit too loud imo.
    Any suggestions for improvement?
    Thanks for the reply, I will give it a try.
    Quote Quote  
  2. Hi there guys, if you download the latest version of Box4 Beta 3 https://www.videohelp.com/software/BOX4 and MKVToolnix https://www.videohelp.com/software/MKVToolNix and the final program Pazera free audio extractor http://www.pazera-software.com/download.php?id=0021&rd=1&ft=p64&f=Pazera_Free_Audio_Ex...t_PORTABLE.zip


    Here is the issues i have noticed with the box4 program, when it normalises the audio within the video files its to high and this causes background noises and sometimes dialogue to fade in and out, i have spent a lot of time trying to config the program and i have noticed that if you drop the frames down to 100 and keep the Guassian filter at the default of 31. also keep the target peak value at 0.90 and maximum gain at 10, this causes the program to do the same but it was not as bad, then i decided to drop the maximum gain right down to 2, audio was perfect apart from i couldn't hear quiet dialogue at times. so i then raised it back up to 10 and noticed that it was slightly fading in and out, like it was clipping but not as bad if i kept the default values. so the problem with this is decibels are far to high when box4 or dynaudnorm normalising the audio to high, what i done was i used mktoolnix to extract the audio then converted each audio file with pazera audio converter but on pazera i had the settings as this,
    Output format MP3 - MPEG-3

    BITRATE MODE = CBR CONSTANT BITRATE

    BITRATE = 192

    SAMPLING FREQUENCY = 48000

    CHANNELS = 2

    now here is where it's going to work guaranteed. on pazera set the volume to 0.22 this will basicly make the audio file very silent, once you have done that get your movie file and insert that converted mp3 back into your movie file, if i was you to make things easier disable all default audio thats within the movie file and just make sure your converted mp3 file is checked, once you have repacked the movie with mktoolnix, open box4 here is the settings i use


    F 100

    G 31

    P 0.90

    m 10

    R 0

    N 0

    C 0

    B 0

    S 0

    Audio ACC 192 KBPS

    Select the D.A.N filter and convert the file and you will have a movie that does not change volume with silent parts and loud parts, it's flawless but if you try and convert the file with the recommended default settings some movies will have some audio parts fading and taking upto 3 seconds to get to a satisfied volume which is anoying. if you go by this guide, this will work for you guarantee'd i spent a lot of time trying to adjust and change the frames with and gaussian filter and i could not get satisfied results, some audio within movie files the dynaudnorm at default setings is making the audio to loud which causes the maximum gain to cause background noises to quickly decrease increase so make sure when you convert to mp3 with pazera to have the volume set at 0.22 and use the settings i have posted above with the box4 movie repacker,

    I'm sure this will just work with dynaudnorm on its own.

    Thanks

    circulationds
    Last edited by circulationds; 14th Oct 2018 at 11:13.
    Quote Quote  
  3. Member
    Join Date
    Oct 2018
    Location
    Gillette Stadium, United States
    Search Comp PM
    Perhaps the Levelator would be a good choice, if you can still find a copy.
    Quote Quote  
  4. Originally Posted by Barrythecrab View Post
    Perhaps the Levelator would be a good choice, if you can still find a copy.

    tried that on movies and i was having that issue as well, audio was rapidly changing in the background, spent about 2 weeks looking for a solution
    Quote Quote  
  5. circulationds,
    I've been using similar settings to yours for the Dynamic Audio Normalizer for quite a while. I use the version built into ffmpeg and mostly convert with foobar2000 as it can be done in a single step. Your process appears to convert to MP3 and then to AAC after normalizing.

    The command line for the converter preset I use with foobar2000 for converting to AAC is the one below. It normalizes with ffmpeg and the Dynamic Audio Normalizer, and then pipes the output to QAAC for conversion to AAC.

    /d /c c:\progra~1\foobar2000\encoders\ffmpeg.exe -i - -ignore_length true -c:a pcm_f32le -af dynaudnorm=f=150 -f wav - | c:\progra~1\foobar2000\encoders\QAAC\qaac.exe --ignorelength -s --no-optimize --no-delay -V 64 -o %d -

    Or for a little more compression I use the following in the command line
    -af dynaudnorm=f=75:g=11
    Like you, I think the default DAN settings respond too slowly.

    https://ffmpeg.org/ffmpeg-all.html#dynaudnorm

    There's also a matrix mixer DSP for fb2k you can add to the conversion chain for downmixing multichannel audio to stereo, and once again it can all be done in a single step. You can save multiple conversion presets. I've posted a few screenshots below. The Matrix Mixer settings I use for downmixing 5.1ch to stereo will drop the volume a fair bit so there's no need to reduce it further before normalizing. For stereo sources you might prefer to reduce the volume before it's normalized, although I don't find it necessary myself. Once again you could save a conversion preset that reduces the volume before it's piped to the encoder.

    Foobar2000 will decode the audio in MKV/MP3/AVI etc, so you can open video files and convert the audio, but if there's chapters they can get in the way so I usually extract the audio with gMKVExtractGUI, convert it with foobar2000 and then replace the old audio with MKVToolNix.

    A DSP for foobar2000 I'd also recommend is called the "ffmpeg Decoder Wrapper". It allows foobar2000 to decode anything ffmpeg can decode. I mostly use it for eac3 as there's no other third party decoder for eac3 I know of.

    Foobar2000 can take a bit of getting used to, and you have to download some encoders/decoders yourself, as fb2k only comes with the ones that can be freely distributed, but if you do a lot of converting I'd recommend you give it a try. It'll also convert multiple files simultaneously or even convert to different formats simultaneously. You just load the files for conversion into a playlist, select some or all of them, right click and choose the appropriate conversion preset.
    Image Attached Thumbnails Click image for larger version

Name:	fb2k.gif
Views:	156
Size:	18.6 KB
ID:	46927  

    Click image for larger version

Name:	mm.gif
Views:	147
Size:	30.1 KB
ID:	46928  

    Click image for larger version

Name:	fb2kencoder.gif
Views:	156
Size:	14.1 KB
ID:	46929  

    Last edited by hello_hello; 15th Oct 2018 at 09:55.
    Quote Quote  
  6. hello_hello how do i use the dynaudnorm on the foobar2000 program?
    Quote Quote  
  7. I use ffmpeg instead of the standalone version, if that's what you're asking, as the standalone version doesn't support a "fake" wave file header. The "fake" header is used to specify things such as the number of channels and sample rate when piping, and because the standalone version doesn't support it, you sometimes have to specify them in the command line, which makes it less friendly to use with a GUI. With a "fake" wave header, that information is automatically passed along to the encoder when the audio is piped. All current encoders support it, as do GUIs, but not the standalone DAN .

    I think the same applies to the input for the DAN when it's being fed raw (decoded) audio, although to get around that you can tell foobar2000 to decode to a temporary wave file first. Then the DAN can compress it and output a sensible wave file.
    I think I'm remembering that correctly. It's been a while since I've used the standalone version. It's easier to use ffmpeg with a GUI.

    Anyway, here's a foobar2000 command line for the DAN itself. "%s" creates a temporary wave file and the DAN would output a wave file. See the screenshot of the encoder configuration below.

    -i %s -o %d -f 75 -g 11

    You can do the same thing with ffmpeg, but there's no need for a temporary input file. This tells ffmpeg to output a 16 bit wave file:

    -i - -ignore_length true -af dynaudnorm=f=75:g=11 -c:a pcm_s16le %d

    Or you can encode with ffmpeg at the same time. It has the LAME mp3 encoder built in. This is the command line I use for converting to VBR MP3, with LAME's default V2 quality (and forcing ID3v2.3 for any tags instead of the default of ID3v2.4). Specify mp3 as the extension and "lossy" as the format in the encoder configuration.

    -i - -ignore_length true -af dynaudnorm=f=150 -c:a libmp3lame -q:a 2 -id3v2_version 3 %d

    Or to encode with ffmpeg's AAC encoder, something like this. Specify m4a or mp4 as the extension and "lossy" as the format in the encoder configuration.

    -i - -ignore_length true -af dynaudnorm=f=150 -c:a aac -vbr 5 %d

    If there's a sub-folder called "encoders" in the foobar2000 folder (it's created if you install the free encoder pack), when you create a custom encoder configuration and specify an exe, foobar2000 will look for it in that folder. You can specify the full path for the encoder if you wish, but if you don't and foobar2000 can't find it in the encoders folder, it'll ask where the exe is located the first time you use the encoder configuration.

    To create a new encoder configuration or converter preset, open at least one audio file, right click on it, select Convert, and in the sub-menu click on the 3 dots at the bottom of the preset list (assuming there's existing presets). That takes you to the converter configuration.

    As a side note... you can open and convert the output of Avisynth scripts containing audio with this DSP: https://foobar2000.com/components/view/foo_input_avs
    Image Attached Thumbnails Click image for larger version

Name:	fb2k dan.gif
Views:	262
Size:	13.8 KB
ID:	46937  

    Click image for larger version

Name:	fb2k ffmpeg.gif
Views:	256
Size:	14.0 KB
ID:	46938  

    Click image for larger version

Name:	fb2k convert.gif
Views:	261
Size:	43.3 KB
ID:	46942  

    Last edited by hello_hello; 15th Oct 2018 at 00:52.
    Quote Quote  
  8. hello_hello

    i see what you mean with getting used to foobar manged to get some movie audio converted but other ones are failing on me, i think i will just use the method i'm using just now but i will use the foobar to normalise them to 89db once box4 has completed, i tried the EBU R128 normalizer through foobar but to me it was like it was the same issue, audio was slightly fading at points but not as bad as the box4 default settings, i have used your settings as a guide with box4 f =75 and the gaussian filter set at 11 so far so good, so thank you for posting, only thing about my way is the conversion with the pazera takes a long time to do, but i do like the foobar2000 as it has some good tweaks i can use for my music library. i was using mp3gain to begin with a few years ago but i did not understand what clipping meant until i started reseasching converting video's and realised that on my dredd movie i heard the clipping sound through my tv so that's what made me look into something that was better for audio conversion. i have the dynaudnorm GUI but its asking me to create a logfile which i'm not entirely sure how to do. i have checked on youtube but i still can't figre it out lol, it's all new to me but enjoying the challenge
    Quote Quote  
  9. If you're referring to the EBU R128 Compressor DSP.... it compresses way too much for me and the window size is a little too big (although it probably conforms to the spec) so sometimes it lets a brief loud peak through, or briefly increases the volume of a quiet section too much.

    For soundtrack audio, ReplayGain's 89dB is often too high (unless you do compress it) as soundtrack audio is more dynamic than CD tracks. I think the R128 standard is -23LUFS (-23dB), which is the same as 84dB for ReplayGain.

    And yeah.... I tried the dynaudnorm GUI at one stage but I remember having a similar issue and I didn't care enough to try to work out what the problem was.

    Have fun!
    Quote Quote  
  10. There's a few really good, interesting, and entriging posts here.

    @circulationds and @hello_hello

    Was interested in reading your experiences with BOX4/DAN and/or FFMPEG/DAN and/or DAN standalone.

    Obviously DAN is going to 'dramatically' change the audio, which of course, is actually usually what a user wants. I find it invaluble. However, that's not to say there hasn't been some things I'd like to change or at least get to the bottom of.

    @circulationds

    Using MP3 was an interesting comment...

    Some time ago, certainly more than a year ago, when testing/using DAN via FFMPEG, to play on our "Smart TV", I use the word "Smart" lightly, I didn't experience any fade in/out problems but did experience 'pops' and 'crackles' now and then (excluding the previous poor AAC encoders). Further testing revealed this would only happen when encoding to AAC. Encoding (with DAN) to MP3 never produced any problem. No problems were ever heard on my PC. These days, on a different TV, all is fine no matter what I do.

    It would be nice to know in which environment, PC, TV, PHONE, you experienced these problems?

    What format is the source audio?

    Maybe it's something we can fix. Is it DAN? Is it the source/output container? Is it dependent on the source/output audio codec? Is it just the settings?

    Can I get any test clips? (video-audio)
    Quote Quote  
  11. hi there video.baba

    I will upload a sample of a clip later tonight that has been untouched with BOX4 default settings and i will post the same clip with the setings i used whilst using pazera, the file that i was having problems with was an AAC 2 channel, it seems to be when there was background noise the dialogue is dramatically lowered and does not change back until 3 seconds roughly, the movie i spent a lot of time with had a few scenes with rain in the background and i spent time working around just that section of the movie, then i did notice that the louder the movie was the more frequent loudness would appear disapear, so i decided to use pazera and lower the actual volume of the converted audio, re-packed it with mktoolnix and tried converting it again with the BOX4 BETA 3 Movie re-packer, and the outcome for that particular part was flawless and same with the entire movie, i then done the same with another file that was fading in and out and again it was flawless, i think it would be good if there was an actual loudness reader on the BOX4 movie repacker so that users had a rough idea how loud the original movie is before using the D.A.N feature.


    I watch movies on my smart tv and i done the editing using headphones so that i could perfect the issues of the fading in and out. As stated above when i lowered the actual Maximum gain factor on BOX4 to 2.0 the audio was not fading in and out using F=100 G=31 P=0.90 M=2.0 R=0 N=0 C=0 B=0 S=0 that's what made my decide to lower the the actual untoched audio file and it worked with maximum gain factor on default level. maybe if there was a feature on BOX4 to lower the actual source file volume before converting and using the D.A.N feature, if not maybe a tips tab on the program advising users to lower main audio if they are experiencing fade in and outs source using a stand alone audio converter, obviously keeping the 2 channel stereo whilst converting.

    It's all new to me as you can see from my replies to @ hello_hello <<<has helped me out a lot and guided me through more options to use but i will post some clips later for you so that you can get a rough idea. but i do have to say i do like the BOX4 beta 3 movie re-packer hats off to you for your hard work

    Cheers
    Last edited by circulationds; 16th Oct 2018 at 11:38.
    Quote Quote  
  12. I only just realised I'd previously posted in this thread. Post #29 has samples attached I created for comparing a few compressors/normalisers.
    Quote Quote  
  13. Here is the samples from the movie i was working on

    Sample 1 is default Box4 3 Beta 2 D.A.N Enabled and selected mp3 format you can clearly hear the dialogue lowering as she is speaking (runtime 50 seconds into the clip) this also happens when i select AAC also

    Sample 2 is Box4 3 Beta 2 D.A.N AAC F 100 G 31 P 0.90 m 10 R 0 N 0 C 0 B 0 S 0 dialogue is a bit better but you can clearly hear the volume pumping louder and quieter if you start to listen you can hear it in the rain it gets louder at points just skip to 3 min 37 and you will hear what i mean,

    sample 3 is Box4 3 Beta 2 exact same setings as sample 2 but before encoding i lowered the volume of the audio file with pareza to 0.22 and the clip is perfect if you listen at 3 min 37 you will hear the rain does not get louder or quieter stays the same maybe an option for the maximum gain needs sorting out on the BOX4 3 Beta 2 maybe lowering it might help, as i said when i use the same setings as sample 2 and lower the maximum gain to just 2.0 its perfect but the issue lies again with not being able to hear lower dialogue very well, it's like the maximum gain filter just over boosts, maybe needs some tweaking.

    Cheers
    Last edited by circulationds; 16th Oct 2018 at 14:55.
    Quote Quote  
  14. circulationds,
    Any chance you could post the original audio too? Or a version converted to a wave file without alteration? If it's multi-channel, could you upload a multichannel version?

    It's not possible to duplicate what you're doing without it and I wouldn't mind trying it with my settings, which aren't far off being the same as yours.

    I can hear what your describing, but I wouldn't mind converting the original to see if the result is the same, although as your samples don't start and end in the same place I found it hard jump to the same point in the audio each time, which made it harder to compare them.

    I really don't think you should have to reduce the volume so much before normalising. Also some of your settings don't match what the DAN command line should be. The way I read it, "-n" would mean "no coupling". There's no "-n 0" setting as such, but Box4 could be translating the options in it's GUI. I was going to take it for a spin, but I see it doesn't play with XP.
    Last edited by hello_hello; 17th Oct 2018 at 06:39.
    Quote Quote  
  15. hello_hello,
    yeh I noticed the samples, thanks, I try get them now.

    circulationds,
    Thanks for the samples, I just had time to download them this morning before work and have a quick listen.
    I did hear what you were referring to with sample 1,but not 2 and 3. However, I didn't use headphones and had little time. I'll try more later.
    The way the different audio samples 'looked' (audio editor) was interesting, so different.

    I agree with hello_hello with the original sample you used, that would be handy. However, I noticed your samples came from an already encoded mp4 file, AAC? I wonder if the same problem would occur from the actual 'real' original - AC3/DTS?
    Quote Quote  
  16. I just had a thought, something I'd love you to try, when you get time of course.

    1. Load your file into BOX4.
    2. Go into SETTINGS, and change the 'LOUDNESS NORMALIZER' option 'I' to '-23' (default is -16).
    3. Click the tracks button of your job and select the audio filter 'LOUDNORM'.
    4. Start (output the file).
    Load that same file back in but this time process it with D.A.N (default settings).
    What's the output audio like?
    Quote Quote  
  17. circulationds first sample sounded fairly typical to me when using the default DAN settings. The default Gaussian filter window size is 31 and the help file says:
    "the default value of 31 takes into account the current frame, as well as the 15 preceding frames and the 15 subsequent frames".
    By default the frame size is 500ms, so while I don't know exactly how it's all calculated, it's a pretty big window, making it somewhat slow to respond.

    Post #29 in this thread contains some samples I uploaded a while ago, one of them is compressed with the DAN and nothing but -f 150 in the command line. The source audio is also attached. Maybe circulations will try that one to see how the result compares?

    I don't use the DAN much for my own encoding as I use my PC to watch video, so I compress the audio on playback with a Winamp plugin courtesy of ffdshow. It's options have different names but it works much like the DAN.

    I did learn to never include the LFE channel when downmixing and compressing. It almost never contains anything that isn't present in the stereo channels as it's just supposed to add to what's already there. I found constant low frequencies (such as background music with lots of bass) could cause the normaliser to reduce the amount by which it would otherwise increase the volume. The end result was "foreground speech" sometimes dropped when there was a lot of low frequency stuff in the background. Downmixing without the LFE channel makes that less likely to happen.
    Quote Quote  
  18. Originally Posted by video.baba View Post
    I just had a thought, something I'd love you to try, when you get time of course.

    1. Load your file into BOX4.
    2. Go into SETTINGS, and change the 'LOUDNESS NORMALIZER' option 'I' to '-23' (default is -16).
    3. Click the tracks button of your job and select the audio filter 'LOUDNORM'.
    4. Start (output the file).
    Load that same file back in but this time process it with D.A.N (default settings).
    What's the output audio like?
    Just converted it with the above setings you recommended with loudnorm then converted that file with the D.A.N default setings and it still does that same but its not as low
    Quote Quote  
  19. Originally Posted by hello_hello View Post
    circulationds,
    Any chance you could post the original audio too? Or a version converted to a wave file without alteration? If it's multi-channel, could you upload a multichannel version?

    It's not possible to duplicate what you're doing without it and I wouldn't mind trying it with my settings, which aren't far off being the same as yours.

    I can hear what your describing, but I wouldn't mind converting the original to see if the result is the same, although as your samples don't start and end in the same place I found it hard jump to the same point in the audio each time, which made it harder to compare them.

    I really don't think you should have to reduce the volume so much before normalising. Also some of your settings don't match what the DAN command line should be. The way I read it, "-n" would mean "no coupling". There's no "-n 0" setting as such, but Box4 could be translating the options in it's GUI. I was going to take it for a spin, but I see it doesn't play with XP.

    Here is the 2 original samples you have requested, in wav form have tried both of these with the default D.A.N settings and both do the exact same when converting with BOX4

    AAC 2 Channel

    AC-3 6 Channel
    Image Attached Files
    Last edited by circulationds; 17th Oct 2018 at 09:35.
    Quote Quote  
  20. I meant the 6 channel version without it being downmixed, but not to worry.

    According to an R128 scan, the volume of the AAC version is -18.54 LUFS with a +0.10 dB peak. The AC3 version is -23.53 LUFS with a -1.50 dB Peak. So using the scan as a guide, the AC3 version might have a few more dB of dynamic range, or maybe the peaks of the AAC version were clipped because it's louder.

    Anyway.... I've attached some normalised samples of the AC3 version. I normalised while converting to wave (using the CLI version of the DAN and foobar2000), scanned the output wave files to save their volumes, then converted each to AAC while adjusting their overall volumes to -23 dB. It might make it a bit easier to compare them to the original that way. Before I adjusted the volumes, they ranged from -12.92 dB to -15.71 dB and they all had (true) peaks of +0.17 dB.
    I also threw in a version normalised with the RockSteady Winamp plugin I use on playback. I was curious to see how it'd compare.

    I can't listen too closely at the moment as it's late and I don't have headphones, but from what I've heard I don't think the default DAN settings did too bad a job.
    "-f 150" is probably close to the best compromise. "-f 150 -g 11" and "-f 75 -g 11" were into volume pumping territory at he end. I guess the window size is too small so the volume of the rain sometimes increased and decreased noticeably between the speech.

    There's probably no true "one size fits all" normaliser setting, but you should be able to find a setting that works well most of the time. It could be worse though. If you had to use a traditional compressor and reduce the volume above a certain threshold, you'd probably have to adjust it for each input file and run trial and error encodes. At least with the "increase the volume of the quiet parts" method, the volume of the input file doesn't seem to be anywhere near as critical.

    Code:
    DAN defaults              -23.04 LUFS,	peak -6.94 dB
    DAN -f 150                 -23.04 LUFS,	peak -7.90 dB
    DAN -f 150 -g 11        -23.03 LUFS,	peak -8.93 dB
    DAN -f 75 -g 11          -23.04 LUFS,	peak -9.71 dB
    Rock Steady               -23.04 LUFS,	peak -7.15 dB
    Image Attached Files
    Last edited by hello_hello; 18th Oct 2018 at 00:58.
    Quote Quote  
  21. thanks hello_hello, sorry about the 6 channel wasn't sure how to do that, the samples you posted i think 2 and 5 are the best
    Quote Quote  
  22. So can you produce the same result with Box4, and more importantly, are you happy with it?
    Quote Quote  
  23. not with box4 but good results with foobar, yeah very happy with it thanks for all your help
    Last edited by circulationds; 18th Oct 2018 at 03:25.
    Quote Quote  
  24. Originally Posted by hello_hello View Post
    I meant the 6 channel version without it being downmixed, but not to worry.

    According to an R128 scan, the volume of the AAC version is -18.54 LUFS with a +0.10 dB peak. The AC3 version is -23.53 LUFS with a -1.50 dB Peak. So using the scan as a guide, the AC3 version might have a few more dB of dynamic range, or maybe the peaks of the AAC version were clipped because it's louder.

    Anyway.... I've attached some normalised samples of the AC3 version. I normalised while converting to wave (using the CLI version of the DAN and foobar2000), scanned the output wave files to save their volumes, then converted each to AAC while adjusting their overall volumes to -23 dB. It might make it a bit easier to compare them to the original that way. Before I adjusted the volumes, they ranged from -12.92 dB to -15.71 dB and they all had (true) peaks of +0.17 dB.
    I also threw in a version normalised with the RockSteady Winamp plugin I use on playback. I was curious to see how it'd compare.

    I can't listen too closely at the moment as it's late and I don't have headphones, but from what I've heard I don't think the default DAN settings did too bad a job.
    "-f 150" is probably close to the best compromise. "-f 150 -g 11" and "-f 75 -g 11" were into volume pumping territory at he end. I guess the window size is too small so the volume of the rain sometimes increased and decreased noticeably between the speech.

    There's probably no true "one size fits all" normaliser setting, but you should be able to find a setting that works well most of the time. It could be worse though. If you had to use a traditional compressor and reduce the volume above a certain threshold, you'd probably have to adjust it for each input file and run trial and error encodes. At least with the "increase the volume of the quiet parts" method, the volume of the input file doesn't seem to be anywhere near as critical.

    Code:
    DAN defaults              -23.04 LUFS,	peak -6.94 dB
    DAN -f 150                 -23.04 LUFS,	peak -7.90 dB
    DAN -f 150 -g 11        -23.03 LUFS,	peak -8.93 dB
    DAN -f 75 -g 11          -23.04 LUFS,	peak -9.71 dB
    Rock Steady               -23.04 LUFS,	peak -7.15 dB

    you're a legend m8 thank you so much for your help i'm using the mp3 format i think it sounds better -i - -ignore_length true -af dynaudnorm=f=150 -c:a libmp3lame -q:a 2 -id3v2_version 3 %d ripped about 100 files so far and they all sound the same, i have it set at 83 decibels through foobar i have like 1200 movies to work through and it was doing my head in having some that where fading in and out, it's mainly used for my plex that i stream onto my smart tv, again thank you so much for all the help
    Quote Quote  
  25. Originally Posted by circulationds View Post
    you're a legend m8 thank you so much for your help i'm using the mp3 format i think it sounds better -i - -ignore_length true -af dynaudnorm=f=150 -c:a libmp3lame -q:a 2 -id3v2_version 3 %d ripped about 100 files so far and they all sound the same, i have it set at 83 decibels through foobar i have like 1200 movies to work through and it was doing my head in having some that where fading in and out, it's mainly used for my plex that i stream onto my smart tv, again thank you so much for all the help
    I'm glad you're happy, but.....

    How have you set it to 83 dB?
    After the source is decoded, the volume is adjusted by the ReplayGain section under Processing (if it's enabled in the converter setup), processed by any DSPs in the chain, and then piped to the encoder.

    For the ReplayGain section to adjust the level to a target volume, the source files have to be scanned and the ReplayGain info saved. Normally you'd probably save TrackGain info, so to adjust to 83dB you'd select "Track" as the Source Mode in the ReplayGain section, "Apply Gain" as the Processing option, then adjust the "With RG Info" preamp to -6dB. That'd result in the volume being adjusted to 83 dB in ReplayGain-speak.

    The problem is though, if any DSPs effect the volume you'll no longer have 83 dB, and that would include downmixing multichannel audio to stereo. The same applies to an encoder adjusting the volume, which would happen when ffmpeg is normalising with the dynaudnorm. It can't work any other way because there's no way for fb2k to know what the final volume will be if it's adjusted by a DSP or encoder.

    If you want 83 dB you can do it two ways. One is to make it a 2 pass process. First encode to a lossless file while normalising, then scan the output files and save the ReplayGain data,. From there you can encode to the final format while adjusting the volume to 83dB.
    Alternatively, you can encode as MP3 or AAC and adjust the volume of the output files losslessly as you would with MP3Gain. You'd scan the normalised MP3 or AAC output files, save the ReplayGain data, then under the right click ReplayGain option, select "Apply track ReplayGain to file content".

    Personally I wouldn't bother adjusting the volume for movies after normalising as it's not really important that they're all the same volume as such. For related audio, such as episodes of a TV show it might be better to have them all the same volume, although after normalising they probably won't differ all that much.

    While I'm thinking of it, I'm pretty sure fb2k is set to adjust the volume of files containing ReplayGain data on playback by default. That was the idea of ReplayGain in the first place, but it means you're not always hearing audio at the "real" volume.
    There's options in Preferences under "Playback" where you can disable the ReplayGain adjustment while playing audio. Set the Source Mode and Processing options to "none".

    In case you haven't found them yet there's a few options under Tools/ReplayGain for configuring the ReplayGain scanner, and under "Alter File Content" you can set the target volume for adjusting MP3 and AAC losslessly. And one other little tweak as I think it's disabled by default...

    Under "Advanced/Tools/ReplayGain Scanner" there's an option labelled "Results Dialogue: Advanced formatting of peak values". Check it if it's not checked by default. Translated to meaningful English, it means the peak values displayed by fb2k after a ReplayGain scan will be shown as both decibels and a percentage. Without that option checked the peak values are show as a percentage only. I think it's done that way to conform to the original ReplayGain spec, but percentages don't translate to dB in an intuitive way.

    Anyway, I just thought I'd explain how the volume adjustment works in case you encode 100s of files thinking the volume was being adjusted when it wasn't.

    Have fun.
    Last edited by hello_hello; 18th Oct 2018 at 20:30.
    Quote Quote  
  26. Originally Posted by hello_hello View Post
    Originally Posted by circulationds View Post
    you're a legend m8 thank you so much for your help i'm using the mp3 format i think it sounds better -i - -ignore_length true -af dynaudnorm=f=150 -c:a libmp3lame -q:a 2 -id3v2_version 3 %d ripped about 100 files so far and they all sound the same, i have it set at 83 decibels through foobar i have like 1200 movies to work through and it was doing my head in having some that where fading in and out, it's mainly used for my plex that i stream onto my smart tv, again thank you so much for all the help
    I'm glad you're happy, but.....

    How have you set it to 83 dB?
    After the source is decoded, the volume is adjusted by the ReplayGain section under Processing (if it's enabled in the converter setup), processed by any DSPs in the chain, and then piped to the encoder.

    For the ReplayGain section to adjust the level to a target volume, the source files have to be scanned and the ReplayGain info saved. Normally you'd probably save TrackGain info, so to adjust to 83dB you'd select "Track" as the Source Mode in the ReplayGain section, "Apply Gain" as the Processing option, then adjust the "With RG Info" preamp to -6dB. That'd result in the volume being adjusted to 83 dB in ReplayGain-speak.

    The problem is though, if any DSPs effect the volume you'll no longer have 83 dB, and that would include downmixing multichannel audio to stereo. The same applies to an encoder adjusting the volume, which would happen when ffmpeg is normalising with the dynaudnorm. It can't work any other way because there's no way for fb2k to know what the final volume will be if it's adjusted by a DSP or encoder.

    If you want 83 dB you can do it two ways. One is to make it a 2 pass process. First encode to a lossless file while normalising, then scan the output files and save the ReplayGain data,. From there you can encode to the final format while adjusting the volume to 83dB.
    Alternatively, you can encode as MP3 or AAC and adjust the volume of the output files losslessly as you would with MP3Gain. You'd scan the normalised MP3 or AAC output files, save the ReplayGain data, then under the right click ReplayGain option, select "Apply track ReplayGain to file content".

    Personally I wouldn't bother adjusting the volume for movies after normalising as it's not really important that they're all the same volume as such. For related audio, such as episodes of a TV show it might be better to have them all the same volume, although after normalising they probably won't differ all that much.

    While I'm thinking of it, I'm pretty sure fb2k is set to adjust the volume of files containing ReplayGain data on playback by default. That was the idea of ReplayGain in the first place, but it means you're not always hearing audio at the "real" volume.
    There's options in Preferences under "Playback" where you can disable the ReplayGain adjustment while playing audio. Set the Source Mode and Processing options to "none".

    In case you haven't found them yet there's a few options under Tools/ReplayGain for configuring the ReplayGain scanner, and under "Alter File Content" you can set the target volume for adjusting MP3 and AAC losslessly. And one other little tweak as I think it's disabled by default...

    Under "Advanced/Tools/ReplayGain Scanner" there's an option labelled "Results Dialogue: Advanced formatting of peak values". Check it if it's not checked by default. Translated to meaningful English, it means the peak values displayed by fb2k after a ReplayGain scan will be shown as both decibels and a percentage. Without that option checked the peak values are show as a percentage only. I think it's done that way to conform to the original ReplayGain spec, but percentages don't translate to dB in an intuitive way.

    Anyway, I just thought I'd explain how the volume adjustment works in case you encode 100s of files thinking the volume was being adjusted when it wasn't.

    Have fun.
    Hi Hello_Hello i'm now using the rocksteady plugin with foobar i have used your settings and wow movies sound way better than using dynaudnorm, i'm still getting to grips on what's what on it. One problem i'm having though is when i save the settings on the rocksteady it keeps resetting and i have saved the presets, do i have to set them presets every time i want to convert the audio? i mean i have saved a preset list and for some reason it keeps resetting, i have also installed the free codec pack so that i can convert to lossey files but when i select the format i'm getting N/A and foobar tells me to install the codec pack which i have?
    Quote Quote  
  27. I don't think you can save a preset as such with the RockSteady configuration. Just adjust the settings and click okay. Or maybe click "apply" first, but they should "stick".

    I'm not sure what's happening with the conversion to a lossy format. Which one? After a conversion fails, open the Console from View/Console and see what it says there. For converting to mp3, I often use ffmpeg instead of the LAME exe. FFmpeg accepts a 32 bit float input. LAME does also these days (from memory) but it converts the input to 24 bit integer before encoding which increases the potential for clipped peaks. I don't think ffmpeg does that.

    -i - -ignore_length true -c:a libmp3lame -q:a 2 -id3v2_version 3 %d

    The encoders should be installed to a sub-folder called "encoders" in the foobar2000 folder. Maybe post a screenshot of your encoder configuration to see if something's wrong there. Even though fb2k has AAC presets, I think they all require you to download certain files yourself. The other encoders should work with the free encoder pack installed though.

    Free_Encoder_Pack_2018-05-22.exe (or whatever the current version is) can be unzipped like a zip file with 7-zip. Then you could manually copy the encoders to the encoders folder if need be, but installing should work, unless they're being installed to the wrong location for some reason.

    I have a file called "plugin.ini" in the same folder as RockSteady. It's just a text file with an "ini" extension. Here's the contents. Use Notepad to create and save the ini file if it doesn't exist and put it in the folder where RockSteady is located. Normally "C:\Program Files\Winamp\Plugins" (for XP) even if WinAmp isn't installed. Hopefully that'll make the settings stick.

    [Piettro Pro's RockSteady]
    FullAmpTo=30
    MaxAmpDB=15
    AmpGainTime=10
    PresetNo=-1
    SmartLimit=1
    AmplifyTo=-1
    RelPos=100
    JointStereo=1

    For RockSteady, the help file says everything below the "Amplify up to a level of" value is amplified and anything above it is left alone. The "Maximum amplification" setting is the maximum amount anything below the "Full amplification up to" value can be amplified. Above the "Full amplification up to" value, the help file says "the amplification amount will be just linearly interpolated up to the maximum level".
    I've been using 30% for the "Full amplification up to" value lately, which is -10.45dB and 25% is -12.04dB, so there's not much difference. Just a chance for a slight increase in any "volume pumping" of fairly quiet sounds at 30% (I assume). Although that's -10.45dB or -12.04dB if the "Amplify up to a level of" value was set to 0dB, as it's a percentage of the "Amplify up to a level of" volume. I'd have to think about the math when it's a percentage of a lower volume.

    With the "Amplify up to a level of" setting at -1dB, pretty much everything is amplified to some degree, but there's a chance of mild clipping (nothing I've heard) even with the "smart limit" option enabled. If you're fussed about that, you might want to reduce the "Amplify up to a level of" value a bit, and/or tell fb2k to reduce the volume a little to allow for some more headroom. Rather than use the ReplayGain section for that, you can put the Amplify DSP ahead of RockSteady in the DSP chain and reduce the volume that way. It's probably easier. -2dB or -3dB should be plenty. https://foobar.hyv.fi/?view=foo_dsp_amp

    (Edited to correct a mistake in the way I explained the RockSteady settings above)
    Last edited by hello_hello; 29th Oct 2018 at 08:10.
    Quote Quote  
  28. Originally Posted by hello_hello View Post
    I don't think you can save a preset as such with the RockSteady configuration. Just adjust the settings and click okay. Or maybe click "apply" first, but they should "stick".

    I'm not sure what's happening with the conversion to a lossy format. Which one? After a conversion fails, open the Console from View/Console and see what it says there. For converting to mp3, I often use ffmpeg instead of the LAME exe. FFmpeg accepts a 32 bit float input. LAME does also these days (from memory) but it converts the input to 24 bit integer before encoding which increases the potential for clipped peaks. I don't think ffmpeg does that.

    -i - -ignore_length true -c:a libmp3lame -q:a 2 -id3v2_version 3 %d

    The encoders should be installed to a sub-folder called "encoders" in the foobar2000 folder. Maybe post a screenshot of your encoder configuration to see if something's wrong there. Even though fb2k has AAC presets, I think they all require you to download certain files yourself. The other encoders should work with the free encoder pack installed though.

    Free_Encoder_Pack_2018-05-22.exe (or whatever the current version is) can be unzipped like a zip file with 7-zip. Then you could manually copy the encoders to the encoders folder if need be, but installing should work, unless they're being installed to the wrong location for some reason.

    I have a file called "plugin.ini" in the same folder as RockSteady. It's just a text file with an "ini" extension. Here's the contents. Use Notepad to create and save the ini file if it doesn't exist and put it in the folder where RockSteady is located. Normally "C:\Program Files\Winamp\Plugins" (for XP) even if WinAmp isn't installed. Hopefully that'll make the settings stick.

    [Piettro Pro's RockSteady]
    FullAmpTo=30
    MaxAmpDB=15
    AmpGainTime=10
    PresetNo=-1
    SmartLimit=1
    AmplifyTo=-1
    RelPos=100
    JointStereo=1

    For RockSteady, the help file says everything below the "Amplify up to a level of" value is amplified and anything above it is left alone. The "Maximum amplification" setting is the maximum amount anything below the "Full amplification up to" value can be amplified. Above the "Full amplification up to" value, the help file says "the amplification amount will be just linearly interpolated up to the maximum level".
    I've been using 30% for the "Full amplification up to" value lately, which is -10.45dB and 25% is -12.04dB, so there's not much difference. Just a chance for a slight increase in any "volume pumping" of fairly quiet sounds at 30% (I assume). Although that's -10.45dB or -12.04dB if the "Amplify up to a level of" value was set to 0dB, as it's a percentage of the "Amplify up to a level of" volume. I'd have to think about the math when it's a percentage of a lower volume.

    With the "Amplify up to a level of" setting at -1dB, pretty much everything is amplified to some degree, but there's a chance of mild clipping (nothing I've heard) even with the "smart limit" option enabled. If you're fussed about that, you might want to reduce the "Amplify up to a level of" value a bit, and/or tell fb2k to reduce the volume a little to allow for some more headroom. Rather than use the ReplayGain section for that, you can put the Amplify DSP ahead of RockSteady in the DSP chain and reduce the volume that way. It's probably easier. -2dB or -3dB should be plenty. https://foobar.hyv.fi/?view=foo_dsp_amp

    (Edited to correct a mistake in the way I explained the RockSteady settings above)

    Hi mate i had to uninstall foobar and reinstall it, i'm having issues adding the rocksteady dsp again its not showing up in the dsp list, any ideas?
    Quote Quote  
  29. Not really, aside from asking if you installed the Winamp DSP bridge again? I assume you installed it the first time around? It's the Winamp bridge that shows up in the DSP list.

    Something I didn't know, or I'd forgotten..... the Winamp DSP bridge limits the peaks automatically when it converts from floating point to fixed point for the Winamp plugins. It's mentioned here. According to the info in the link below, reducing the level by 6dB before sending audio to the Winamp DSP bridge will prevent it limiting. It sounds like it includes something like fb2k's -6dB Hard Limiter DSP. I guess I didn't notice as I only use it to compress with RockSteady anyway. Plus mostly I'd be downmixing with the matrix mixer first, which reduces the volume a fair bit to prevent clipping when it's normalise option is enabled.

    https://hydrogenaud.io/index.php/topic,49356.msg749713.html#msg749713
    Last edited by hello_hello; 29th Oct 2018 at 17:17.
    Quote Quote  
  30. Originally Posted by hello_hello View Post
    Not really, aside from asking if you installed the Winamp DSP bridge again? I assume you installed it the first time around? It's the Winamp bridge that shows up in the DSP list.

    Something I didn't know, or I'd forgotten..... the Winamp DSP bridge limits the peaks automatically when it converts from floating point to fixed point for the Winamp plugins. It's mentioned here. According to the info in the link below, reducing the level by 6dB before sending audio to the Winamp DSP bridge will prevent it limiting. It sounds like it includes something like fb2k's -6dB Hard Limiter DSP. I guess I didn't notice as I only use it to compress with RockSteady anyway. Plus mostly I'd be downmixing with the matrix mixer first, which reduces the volume a fair bit to prevent clipping when it's normalise option is enabled.

    https://hydrogenaud.io/index.php/topic,49356.msg749713.html#msg749713

    Hi mate, read up on your reply, i have moved onto another option, i wasnt happy with the outcome of winamp dsp bridge, some movies sounded like they were compressed to much, i have used in my active dsp list whilst converting as follows and the outcome has been excellent! don't ask me how but this little dsp really works its called Vlevel, i'm using downmix to stereo then followed by your matrix mixer setings. I know that downmixing to stereo, then using the settings on the matrix mixer doesn't make any sense but it seems to be working then i used the vlevel dsp and then i use amplify +4db the results have been really good so far,, and the dynamics sound good as well. (Converting the files to AAC 192 constant bitrate) i have added some pictures, only issue is when i set up a list and after applying gain info i can only convert one file at a time as when i go to batch convert the end audio from the previous film starts in the next movie, its like a fade in fade out and the dialogue is out of sync due to the last few seconds of the previous movie audio is included in the next movie, i just select each file one at a time and convert them that way and do them in batches of 20, but all is working out good, best result i have found upto now
    Image
    [Attachment 47056 - Click to enlarge]
    Image
    [Attachment 47057 - Click to enlarge]
    Last edited by circulationds; 30th Oct 2018 at 16:32.
    Quote Quote  



Similar Threads

Visit our sponsor! Try DVDFab and backup Blu-rays!