Digitizing VHS tapes using MEncoder in Linux

This post is of a high geekitude factor, so be warned. One major reason for writing it is to record what I did so that in future if I need to refresh my memory I can do so quickly. I decided to work on converting all my vhs tapes into digital format. I have quite a number of these tapes lying around, and having them all in digital format should free up plenty of space. The tapes can then go into storage.

To do this, you need the VCR, and a computer with a tv-input card. The input card I use is the AverMedia AverTV Studio. I got this a while back so as to play around with MythTV. This is software that enables you to set up your PC into a Tivo-like device. You might be asking why I did this, seeing that I don't watch any tv. I did it for the challenge. Just so that I could say that I did it and it worked. And work it did. But after sometime, I got tired and uninstalled everything.

Another good thing about the tv tuner card I have is that it can receive FM radio signals, so with the appropriate software I can listen to radio direct from my PC. I haven't done that in a while, since radio has nothing much to offer. So the card had been laying idle in my computer until I came upon the idea of digitizing my vhs tapes.

Apart from the hardware, you will need the software to do the actual processing. Like I keep mentioning, I run Linux, and so the software I use is for this platform, though you can get binaries for Windows. The workhorse software is called MEncoder, which is installed as part of mplayer. Mplayer is a powerful commandline based media player. You can get a graphical user interface (GUI) wrapper for it (such as gmplayer) if you are commandlinephobic.

Before getting things off to a start, you need to set up your sound card to ensure the audio will be captured properly. The settings needed depend on what card you have. Personally, I have the SoundBlaster Live!. The volume settings needed can either be done on the commandline, or using the Alsamixer utility. Either way, make sure the following items are set as shown:

• Master unmuted and set to 100%


• PCM unmuted and set to 75%


• Wave set to 75%


• Line unmuted and set to 75%. For this, you also need to set it as the capture input, meaning the input into which sound from the video will be recorded.


• Capture set to 75%


• AC97 set to 75%. This one is not really necessary for the actual recording, i.e. has no effect on it, but is good to set since it will let you hear the recording. If set to 0, you won't hear anything during the recording.


• AC97 Cap set to 75%. This is necessary, and is a different setting from the AC97.

Once this is done, you are good to go. Set the tape rolling in the VCR, then type the following command on the computer to begin capturing.

mencoder -tv driver=v4l2:width=640:height=480:channel=3:norm=ntsc tv:// -o file.avi -ovc raw -oac pcm

Note that during capture you will not get any video output. You will only get sound (thus the importance of setting the AC97 to 75% for effective feedback).

I now give an explanation of the components of the command typed up above:

-tv signifies options that tune the TV capture module. For this case, the options are:

• driver: This is the software in the computer that interfaces with the actual device which captures the video signal, in this case v4l2.


• width:height: These are the dimensions in pixels of the output recording, in this case, 640 by 480.


• channel: This is the channel the tv tuner card is to be set to, in this case, 3.


• norm: This is the video standard to be used. In this example, NTSC is selected, which is the system in North America.

tv:// means capture the video using the tv capture module. The options listed after this are:

• -o: Indicates what the output file should be named, in this case, file.avi


• -ovc: Indicates the video format to encode to. In this example, raw was selected, meaning the output video should not be compressed.


• -oac: Similar to video, except that this is for audio. In this case pcm is selected. Like raw for video, this is the uncompressed audio format.

Let this run for as long as the tape is running. Once you get to the end or to a point where you want to stop, just type Ctrl-c to stop the recording.

You will now have a file.avi file on your disk. I should point out that captured vidoe files take a huge amount of space on disk. In one capture I did, 40 minutes of video used up 30GB of space. Remember, the command above states that both the video and audio should be uncompressed, thus the huge space needs. Make sure you have enough disk space before carrying out this operation. Make sure you also have enough processor horsepower, as well as lots of memory. I have a 1.1GH AMD processor with 512MB of RAM on my PC. Ancient by the standards of computers out today, but still fast enough for my needs. I don't get any problems recording with this PC. I have 440GB of disk space, more than plenty for my storage needs.

Next you need to give the following long command:

mencoder file.avi -o file.flv -of lavf -oac mp3lame -lameopts abr:br=56 -srate 22050 -ovc lavc -lavcopts vcodec=flv:vbitrate=500:mbd=2:mv0:trell:v4mv:cbp:last_pred=3 -lavfopts i_certify_that_my_video_stream_does_not_use_b_frames

This command basically compresses the output file down to a more manageable size. It also converts the format of the file from one type to another as detailed in the breakdown below:

• file.avi is the input file


• -o indicates what the output file should be. In this case, I type file.flv.


• -of indicates what container output format should be used. in this case, lavf is selected (I won't explain this in layman terms as it would take a whole other post by itself! If you really really want to know what this is, a good place to learn about it can be found here).


• -oac Just as for the capture portion, this indicates what audio format to encode the sound in. For this case, the sound should be encoded to mp3 using lame (an encoding application). Ensure that lame is installed in your system prior to issuing this command, or you will get an error. Do know, however, that you are not limited to lame. You can use other encoders such as bladeenc


• -lameopts indicates what options to use in the lame application during encoding to mp3. In this case, only one option, abr is give. This stands for average bit rate, with the bit rate (br) given as 56, which is in kilobits per second. This means the mp3 file should use 56 kilobits per second of audio. The higher the bit rate, the higher quality the audio, but then again the larger the file.


• -srate indicates the output sample rate to be used in hertz. This is the rate at which the audio is sampled by the soundcard. In this case, 22,050 hz is the rate used.


• -ovc similar to -oac, but for video. The output format is lavc. Again, for details on this, check here


• -lavcopts indicates options for the output video format. These are:
  
  
  
  • vcodec: the output video format to use. In this case, flv is selected (thus the file.flv output file up above). flv is the flash video format. You can choose from any number of formats such as mpeg, wmv, avi, etc
  
  
  • vbitrate: the speed at which encoding should take place in megabits per second. In this case, 500 Mbps is the rate used
  
  
  • mbd, mv0, trell, v4mv, cbp and last_pred: these options are used to indicate more technical aspects of the encoding. If you want the details, you can check out the man page for mplayer in your installation, or online here
  
  
  
  
• -lavfopts: this is an interesting option. the setting given up above, i.e. i_certify_that_my_video_stream_does_not_use_b_frames should be typed as shown, otherwise the encoding won't happen. This is to show that your recording has no b-frames in it, which are predictive frames, meaning they are decoded from previous and next frames in a video clip. The presence of b-frames results in MEncoder not being able to correctly assign timestamps during encoding, leading to broken encodes. Thus the need to have this in place for proper output.

Once this command is done doing it's thing, you will have the flash video file ready for use (or mpg or wmv depending on output video format selected). The reason I chose the flv format is my need to have these video files present online. I wanted to play them using the flash video player which only accepts flv files. In order to do this, you will need to have a flash video player installed on your site. That is if you want to go that route,

Alternatively, you could always just resort to youtube. In actual fact, the steps I have just described in generating the video file are similar to what youtube does. That is how they are able to get video from users processed into flash video files which they then host on their servers.

A question that might arise is, why go through this pain? Again, like for the mythtv example above, my answer is the same. So I can say I have done it, and in the process learn more about the wonderful use of computers. You might also ask why not get one of them GUI applications that lets me click buttons and make selections then click run and let it take care of everything for me. I could do that, but then again I wouldn't be able to learn the nitty gritty of what is involved in encoding. Doing it on the command line forces me to learn this. Another advantage of the commandline is, automating the process is much easier this way, since I can write up a script that does everything I have described so far and place it on a server somewhere and just feed it input files. Precisely what youtube does. This is one of those moments where you ask yourself why you never thought of it before the youtube guys did. You could have been a gabazillionare by now! Well, you live and learn.

And now I leave you with a clip from the recording I made for illustrative purposes. If you were Strong to Serve or Walked in the Light, and especially so in '94 and '95, and were in the choir to boot, you will most definitely remember this.