Recherche avancée

Sur d’autres sites (7562)

• FFMPEG loudnorm filter does not work in combination with silenceremove filter

  12 mai 2021, par MareikeP

  I want to consistently normalize audio files for TTS model training. The output audio files should meet the following criteria :

  


  

  1. mono channel

  


  2. sample rate of 22050 Hz

  


  3. wav format

  


  4. no silence at beginning and end of audio clip

  


  5. volume of -24 dB

  


  


  I have already fulfilled the first 4 criteria. So far, it works properly.

  


  Normalizing the volume basically works as well with this ffmpeg command -af loudnorm=I=-24:LRA=11:TP=-1.5 , but not in combination with the silence removal : As soon as I remove silence with this ffmpeg command agate=threshold=0.045:attack=0.5:release=500:ratio=5000,silenceremove=start_periods=1:start_threshold=0.0075,areverse,silenceremove=start_periods=1:start_threshold=0.0075,areverse, the loudness normalization does not work any longer : the output volume now varies between -25dB and -32dB instead of the desired -24 dB.

  


  This is the complete ffmpeg command I used :

  


  ffmpeg -i filename.flac -ac 1 -af agate=threshold=0.045:attack=0.5:release=500:ratio=5000,silenceremove=start_periods=1:start_threshold=0.0075,areverse,silenceremove=start_periods=1:start_threshold=0.0075,areverse,loudnorm=I=-24:LRA=11:TP=-1.5,aresample=22050 -y -hide_banner filename.wav


  


  And this is the piece of code that I'm using to run it :

  


  import os

INPUT_DIR = '/home/username/all_data'
OUTPUT_DIR = '/home/username/normalized_data'
for filename in os.listdir(INPUT_DIR):
    wav_filename = filename[:-5] + '.wav'
    command = (f'ffmpeg -i {INPUT_DIR}/{filename} -ac 1 -af agate='
               f'threshold=0.045:attack=0.5:release=500:ratio=5000,'
               f'silenceremove=start_periods=1:start_threshold=0.0075,'
               f'areverse,silenceremove=start_periods=1:start_threshold='
               f'0.0075,areverse,loudnorm=I=-24:LRA=11:TP=-1.5,aresample'
               f'=22050 -y -hide_banner {OUTPUT_DIR}/{wav_filename}')
    os.system(command)


  


  EDIT :

  


  A complete log from the ffmpeg command can be seen here :

  


  username@pop-os:~$ ffmpeg -i /home/username/audios/filename.flac -ac 1 -af agate=threshold=0.045:attack=0.5:release=500:ratio=5000,silenceremove=start_periods=1:start_threshold=0.0075,areverse,silenceremove=start_periods=1:start_threshold=0.0075,areverse,loudnorm=I=-24:LRA=11:TP=-1.5,aresample=22050 /home/username/result.wav
ffmpeg version 4.2.4-1ubuntu0.1 Copyright (c) 2000-2020 the FFmpeg developers
  built with gcc 9 (Ubuntu 9.3.0-10ubuntu2)
  configuration: --prefix=/usr --extra-version=1ubuntu0.1 --toolchain=hardened --libdir=/usr/lib/x86_64-linux-gnu --incdir=/usr/include/x86_64-linux-gnu --arch=amd64 --enable-gpl --disable-stripping --enable-avresample --disable-filter=resample --enable-avisynth --enable-gnutls --enable-ladspa --enable-libaom --enable-libass --enable-libbluray --enable-libbs2b --enable-libcaca --enable-libcdio --enable-libcodec2 --enable-libflite --enable-libfontconfig --enable-libfreetype --enable-libfribidi --enable-libgme --enable-libgsm --enable-libjack --enable-libmp3lame --enable-libmysofa --enable-libopenjpeg --enable-libopenmpt --enable-libopus --enable-libpulse --enable-librsvg --enable-librubberband --enable-libshine --enable-libsnappy --enable-libsoxr --enable-libspeex --enable-libssh --enable-libtheora --enable-libtwolame --enable-libvidstab --enable-libvorbis --enable-libvpx --enable-libwavpack --enable-libwebp --enable-libx265 --enable-libxml2 --enable-libxvid --enable-libzmq --enable-libzvbi --enable-lv2 --enable-omx --enable-openal --enable-opencl --enable-opengl --enable-sdl2 --enable-libdc1394 --enable-libdrm --enable-libiec61883 --enable-nvenc --enable-chromaprint --enable-frei0r --enable-libx264 --enable-shared
  libavutil      56. 31.100 / 56. 31.100
  libavcodec     58. 54.100 / 58. 54.100
  libavformat    58. 29.100 / 58. 29.100
  libavdevice    58.  8.100 / 58.  8.100
  libavfilter     7. 57.100 /  7. 57.100
  libavresample   4.  0.  0 /  4.  0.  0
  libswscale      5.  5.100 /  5.  5.100
  libswresample   3.  5.100 /  3.  5.100
  libpostproc    55.  5.100 / 55.  5.100
Input #0, flac, from '/home/mareike/tts_data/save/audios_flac/0a6c8520-7536-11eb-8338-b7015f354987.flac':
  Duration: 00:00:04.64, start: 0.000000, bitrate: 1090 kb/s
    Stream #0:0: Audio: flac, 44100 Hz, stereo, s32 (24 bit)
Stream mapping:
  Stream #0:0 -> #0:0 (flac (native) -> pcm_s16le (native))
Press [q] to stop, [?] for help
Output #0, wav, to '/home/mareike/result_0a6c8520-7536-11eb-8338-b7015f354987.wav':
  Metadata:
    ISFT            : Lavf58.29.100
    Stream #0:0: Audio: pcm_s16le ([1][0][0][0] / 0x0001), 22050 Hz, mono, s16, 352 kb/s
    Metadata:
      encoder         : Lavc58.54.100 pcm_s16le
size=     138kB time=00:00:03.19 bitrate= 353.0kbits/s speed=14.3x    
video:0kB audio:138kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 0.055375%


  


  Can anyone tell me what I'm doing wrong and how I can finally get the volume normalized to -24 dB (in combination with silence removal) ? Any help is appreciated, thank you very much !

  


• Is there a way to force FFMPEG to decode a video stream with alpha from ​a WebM video encoded with libvpx-vp9 ?

  6 juillet, par David

  I have a ​WebM file with one video stream that was encoded with VP9 (libvpx-vp9).

  


  I wrote a C++ program to extract the frames from the video stream and save them out as PNG's. This works fine except that the resulting PNG's are missing alpha.

  


  If I extract the frames from the same WebM file using FFMPEG the resulting PNG's do contain alpha. Here is the output from FFMPEG :

  


  $ ffmpeg -c:v libvpx-vp9 -i temp/anim.webm temp/output-%3d.png

[libvpx-vp9 @ 0000024732b106c0] v1.10.0-rc1-11-gcb0d8ce31
    Last message repeated 1 times
Input #0, matroska,webm, from 'temp/anim.webm':
  Metadata:
    ENCODER         : Lavf58.45.100
  Duration: 00:00:04.04, start: 0.000000, bitrate: 112 kb/s
  Stream #0:0: Video: vp9 (Profile 0), yuva420p(tv), 640x480, SAR 1:1 DAR 4:3, 25 fps, 25 tbr, 1k tbn, 1k tbc (default)
    Metadata:
      alpha_mode      : 1
      ENCODER         : Lavc58.91.100 libvpx-vp9
      DURATION        : 00:00:04.040000000


  


  FFMPEG identifies the stream format as yuva420p.

  


  Here is the output from my program when av_dump_format is called :

  


  Input #0, matroska,webm, from 'temp/anim.webm':
  Metadata:
    ENCODER         : Lavf58.45.100
  Duration: 00:00:04.04, start: 0.000000, bitrate: 112 kb/s
  Stream #0:0: Video: vp9 (Profile 0), yuv420p(tv), 640x480, SAR 1:1 DAR 4:3, 25 fps, 25 tbr, 1k tbn, 1k tbc (default)
    Metadata:
      alpha_mode      : 1
      ENCODER         : Lavc58.91.100 libvpx-vp9
      DURATION        : 00:00:04.040000000


  


  Notice that the detected stream format is yuv420p (the alpha is missing).

  


  Does anybody know how to force the stream format to use alpha ?

  


  My setup code resembles the following (error handling is omitted)

  


  auto result = avformat_open_input(&formatContext, fileName.c_str(), nullptr, nullptr);
auto result = avformat_find_stream_info(formatContext, nullptr);
streamIndex = av_find_best_stream(formatContext, mediaType, -1, -1, nullptr, 0);
auto stream = formatContext->streams[streamIndex];
const auto codecIdentifier{ AV_CODEC_ID_VP9 };
auto decoder = avcodec_find_decoder(codecIdentifier);
pCodecContext = avcodec_alloc_context3(decoder);
auto result = avcodec_open2(pCodecContext, decoder, &options);
// AV_PIX_FMT_YUV420P - missing alpha
auto pixelFormat = pCodecContext->pix_fmt;


  


  Gyan pointed out what the problem was. Here is the corrected code :

  


  In case anybody else runs into this issue in the future here is the code (error handling omitted) :

  


  auto formatContext = avformat_alloc_context();&#xA;formatContext->video_codec_id = AV_CODEC_ID_VP9;&#xA;const auto decoder = avcodec_find_decoder_by_name("libvpx-vp9");&#xA;formatContext->video_codec = decoder;&#xA;avformat_open_input(&amp;formatContext, fileName.c_str(), nullptr, nullptr);&#xA;avformat_find_stream_info(formatContext.get(), nullptr);&#xA;for (unsigned int streamIndex = 0; streamIndex &lt; formatContext->nb_streams; &#x2B;&#x2B;streamIndex) {&#xA;    // Displayed the stream format as yuva420p (contains alpha)&#xA;    av_dump_format(formatContext, static_cast<int>(streamIndex), fileName.toStdString().c_str(), 0);&#xA;}&#xA;</int>

  


  Thanks,

  


• Feeding raw image bytes into ffmpeg rawvideo fails with Invalid buffer size on linux only

  13 février 2021, par cherouvim

  I have a nodejs program which generates raw (rgb24) image(s), which I then pipe into ffmpeg so it saves as png or mp4. My code looks like this :

  


  const fs = require("fs");
// ...
const outputBuffer = Buffer.alloc(outputPngWidth * 3 * outputPngHeight);
// ... write data into outputBuffer
fs.writeSync(process.stdout.fd, outputBuffer);


  


  I then do the following in CLI :

  


  node generate | ffmpeg -f rawvideo -pixel_format rgb24 -video_size 1000x1000 -i - test.png


  


  Alternatively, if I generate lots of images from my program, I do this to generate the video file :

  


  node generate | ffmpeg -f rawvideo -pixel_format rgb24 -video_size 1000x1000 -r 60 -i - -codec:v libx265 test.mp4


  


  On windows this works flawlessly. On linux (either on Ubuntu 20 VM, or Ubuntu 20 installed directly on a physical machine), it consistently fails with :

  


  pipe:: corrupt input packet in stream 0
[rawvideo @ 0x55f5256c8040] Invalid buffer size, packet size 65536 < expected frame_size 3000000
Error while decoding stream #0:0: Invalid argument


  


  If I split this in 2 phases like so, then it works perfectly on linux as well :

  


  node generate > test.raw
cat test.raw | ffmpeg -f rawvideo -pixel_format rgb24 -video_size 1000x1000 -i - test.png


  


  By looking at the error "packet size 65536 < expected frame_size 3000000" it seems that node's fs.writeSync only sends 65536 bytes at a time, but ffmpeg expects 3000000 bytes (that is 1000 width * 1000 height * 3 channels).

  &#xA;

  If I reduce my image size to something small, e.g 50x50 or 100x100, then it works. As soon as x * y * 3 exceeds 65536, it fails (eg. 160x160 fails with "packet size 65536 < expected frame_size 76800" because 160 * 160 * 3 = 76800).

  &#xA;

  What I've tried so far to solve the issue without luck :

  &#xA;

  &#xA;
  • Force node to spit out the whole buffer at once :

  &#xA;

  &#xA;

  fs.writeSync(process.stdout.fd, outputBuffer, 0, outputBuffer.length);


  &#xA;

  &#xA;
  • All suggestions of Add a big buffer to a pipe between two commands with various linux commands to buffer between node and ffmpeg.

  &#xA;

  • Use Ubuntu 18 instead of 20.

  &#xA;

  • Use node 12 instead of 15.

  &#xA;

  • Figure out a way to change the chunk size in https://nodejs.org/api/fs.html

  &#xA;

  &#xA;

  Is there a way to overcome this ?

  &#xA;