Recherche avancée

Sur d’autres sites (10652)

• Loading and unloading C jni library based on when it's needed

  30 septembre 2014, par Alin

  I finally managed to compile ffmpeg for android and I’ve been able to use it in my app.

  Here is the scenario of my app :

  • I show the user a gridview with thumbnails of images and videos
  • the user can click on a cell and it is taken to image/video details where he can see the full image or play the video
  • the user can apply an image over an video and this is when ffmpeg is used

  So basically, the user might never actually use the watermarking option or he can do it very rare because the amount of available videos is way smaller than images.

  I am loading the ffmpeg library, first time it is needed by running :

  static {
  
       System.loadLibrary("ffmpeglib");
  
  }

  Now here are my questions :

  1. loading the library like this, uses app’s memory and resources ?
  2. can I unload the library, or better said, is it needed to unload it ? I have not found any java code like System.unloadLibrary to take care of unloading
  3. Since the library might be used rarely, wouldn’t a load => do encoding => unload be a better approach ? Or maybe having it loaded would allow easy reuse since no loading is necessary.
  4. If I use an IntentService to load the library and make the encoding, when the service completes the job, does the library gets unloaded ?

• Running multiple services in a Docker container

  27 octobre 2015, par andrea-f

  I created a Docker image with Ubuntu 14.04 and compiled FFMPEG to run the streaming of a video asset to a DASH endpoint. On the same image I can run the media analysis script which basically use FFMPEG and other tools to analyse a video asset. Now I want to put a Django app so that assets can be both loaded in the streaming pipeline and run through the media analysis. What would you suggest is the best approach ? Have 2 Docker images – one with compiled FFMPEG and the streaming pipeline and another one with django, and then share the code between the two ? Or just keep 1 docker image and run both the FFMPEG streaming pipeline and media analysis and Django from there ?

  I am open to suggestions…

  Possible duplicate of http://serverfault.com/questions/706736/sharing-code-base-between-docker-containers

• librosa can't load wav file in aws lambda docker

  30 novembre 2022, par Luka Savic

  I have an AWS Lambda function created using Docker.
I have librosa installed, ffmpeg installed using the solution from this question : install ffmpeg on amazon ecr linux python

  


  I checked in a Lambda function with os.system("ffmpeg -version") and I managed to get valid output, stating different versions and parts of ffmpeg.

  


  Problem is that when I do librosa.load(wav_file) it gives the following error :

  


  /your/path/.venv/lib/python3.9/site-packages/librosa/util/decorators.py:88: UserWarning: PySoundFile failed. Trying audioread instead.
  return f(*args, **kwargs) 


  


  From what I've read, librosa should natively support .wav files, even without ffmpeg, and even though I have ffmpeg installed, it doesn't work.

  


  One more information, .wav file was downloaded, player, and loaded with librosa on my local PC without any problems. I tried also on different wav and mp3 files, and the problems were still there.