Recherche avancée

Sur d’autres sites (4370)

• FFMpeg : audio resampling changes slightly the speed of the music [on hold]

  24 octobre 2018, par thenaoh

  In my Android app, I use the FFMpeg library to extract audio samples from an audio file on the fly, and also resample them (since my file is encoded in 44100 Hz, and my device expects 48000 Hz sound).

  The problem is : even if the sound quality is perfect, the speed is a little bit higher than the regular speed of the music.

  The difference is very small (small enough so you won’t hear it), but you can hear it when you play the same song at the same time on a regular player (like VLC for example).

  I think it comes from the resampling, but I don’t know how to fix it.

  Here is my code (I get my samples thanks to the getPcmFloat() function) : https://gist.github.com/mregnauld/2538d98308ad57eb75cfcd36aab5099a

  How can I correct the speed of the music ?

  Thanks for your help.

• Creating a sequence of images from lyrics to use in ffmpeg

  19 septembre 2018, par SKS

  I’m trying to make an MP3 + Lyric -> MP4 program in python.

  I have a lyrics file like this :

  [00:00.60]Revelation, chapter 4
  
  [00:02.34]After these things I looked, 
  
  [00:04.10]and behold a door was opened in heaven, 
  
  [00:06.41]and the first voice which I heard, as it were, 
  
  [00:08.78]of a trumpet speaking with me, said: 
  
  [00:11.09]Come up hither, 
  
  [00:12.16]and I will shew thee the things which must be done hereafter.
  
  [00:15.78]And immediately I was in the spirit: 
  
  [00:18.03]and behold there was a throne set in heaven, 
  
  [00:20.72]and upon the throne one sitting.
  
  [00:22.85]And he that sat, 
  
  [00:23.91]was to the sight like the jasper and the sardine stone; 
  
  [00:26.97]and there was a rainbow round about the throne, 
  
  [00:29.16]in sight like unto an emerald.
  
  [00:31.35]And round about the throne were four and twenty seats; 
  
  [00:34.85]and upon the seats, four and twenty ancients sitting, 
  
  [00:38.03]clothed in white garments, and on their heads were crowns of gold.
  
  [00:41.97]And from the throne proceeded lightnings, and voices, and thunders; 
  
  [00:46.03]and there were seven lamps burning before the throne, 
  
  [00:48.60]which are the seven spirits of God. 
  
  [00:51.23]And in the sight of the throne was, as it were, 
  
  [00:53.79]a sea of glass like to crystal; 
  
  [00:56.16]and in the midst of the throne, and round about the throne, 
  
  [00:59.29]were four living creatures, full of eyes before and behind.
  
  [01:03.79]And the first living creature was like a lion: 

  I’m trying to create a sequence of images from the lyrics to use into ffmpeg.

  os.system(ffmpeg_path + " -r 2 -i " + images_path + "image%1d.png -i " + audio_file + " -vcodec mpeg4 -y " + video_name)

  I tried finding out the number of images to make for each line. I’ve tried subtracting the seconds of the next line from the current line. It works but produces very inconsistent results.

  import os
  
  import datetime
  
  import time
  
  import math
  
  from PIL import Image, ImageDraw
  
  
  
  
  
  ffmpeg_path = os.getcwd() + "\\ffmpeg\\bin\\ffmpeg.exe"
  
  images_path = os.getcwd() + "\\test_output\\"
  
  audio_file = os.getcwd() + "\\audio.mp3"
  
  lyric_file = os.getcwd() + "\\lyric.lrc"
  
  
  
  video_name = "movie.mp4"
  
  
  
  
  
  def save():
  
  
  
      lyric_to_images()
  
      os.system(ffmpeg_path + " -r 2 -i " + images_path + "image%1d.png -i " + audio_file + " -vcodec mpeg4 -y " + video_name)
  
  
  
  
  
  def lyric_to_images():
  
  
  
      file  = open(lyric_file, "r")
  
  
  
      data = file.readlines()
  
  
  
      startOfLyric = True
  
      lstTimestamp = []
  
  
  
      images_to_make = 0
  
      from_second = 0.0
  
      to_second = 0.0
  
  
  
      for line in data:
  
          vTime = line[1:9] # 00:00.60
  
  
  
          temp = vTime.split(':')
  
  
  
          minute = float(temp[0])
  
          #a = float(temp[1].split('.'))
  
          #second = float((minute * 60) + int(a[0]))
  
          second = (minute * 60) + float(temp[1])
  
  
  
          lstTimestamp.append(second)
  
  
  
      counter = 1
  
  
  
      for i, second in enumerate(lstTimestamp):
  
  
  
          if startOfLyric is True:
  
              startOfLyric = False
  
              #first line is always 3 seconds (images to make = 3x2)
  
              for x in range(1, 7):
  
                  writeImage(data[i][10:], 'image' + str(counter))
  
                  counter += 1
  
          else:
  
              from_second = lstTimestamp[i-1]
  
              to_second = second
  
  
  
              difference = to_second - from_second
  
              images_to_make = int(difference * 2)
  
  
  
              for x in range(1, int(images_to_make+1)):
  
                  writeImage(data[i-1][10:], 'image'+str(counter))
  
                  counter += 1
  
  
  
      file.close()
  
  
  
  def writeImage(v_text, filename):
  
  
  
      img = Image.new('RGB', (480, 320), color = (73, 109, 137))
  
  
  
      d = ImageDraw.Draw(img)
  
      d.text((10,10), v_text, fill=(255,255,0))
  
  
  
      img.save(os.getcwd() + "\\test_output\\" + filename + ".png")
  
  
  
  
  
  save()

  Is there any efficient and accurate way to calculate how many images I need to create for each line ?

  Note : Whatever many images I create will have to be multiplied by 2 because I’m using -r 2 for FFmpeg (2 FPS).

• Extract raw audio frames from OGG music file with Android NDK

  31 octobre 2018, par thenaoh

  In my Android app, I would like to be able to process audio on the fly from an OGG file by extracting audio samples, process them and redirect them to the audio output.

  I know how to make the last 2 steps using Android NDK, but I don’t know how to extract audio samples to get them in an array of floats or shorts.

  I tried to make this code work that, apparently, can extract raw audio samples on the fly.

  The problem is : I don’t manage to add FFMpeg in my project. I tried many tutorials (like this one), but it seems pretty difficult since I work on Windows. So after a while, I found Prebuild FFMpeg for Android, that seems interesting since it’s available for armeabi-v7a, arm64-v8a, x86 and x86_64 architectures, but again, I don’t understand how to add it in my project.

  I also took a look at libogg, libvorbis and vorbisfile, but I have no idea how to add them in my project.

  So, does anyone have a working example on how to extract audio samples from an OGG file on the fly ?

  Thanks for your help.