Loading ...
Sorry, an error occurred while loading the content.

Re: [podcasters] Dynamic Ad Insertion

Expand Messages
  • Stephen Eley
    ... Well, that was stupid. Oh well. If anyone wants my number or Skype handle, I guess you have it now. No big deal. Good thing I didn t have a reason to
    Message 1 of 11 , Aug 1, 2006
    • 0 Attachment
      On 8/1/06, Stephen Eley <sfeley@...> wrote:
      >
      > I just made a public reply to you on the list, but I wanted to put a
      > hand out privately as well and let you know that what you're asking
      > for isn't really all that hard.

      Well, that was stupid.

      Oh well. If anyone wants my number or Skype handle, I guess you have
      it now. No big deal. Good thing I didn't have a reason to e-mail him
      my Social Security number!


      --
      Have Fun,
      Steve Eley (sfeley@...)
      ESCAPE POD - The Science Fiction Podcast Magazine
      http://www.escapepod.org
    • robert
      ... Eric - for item number 1. Me and Mur cover what the splits are for many of the Automatic Ad Insertion networks, in our book Tricks of the Podcasting
      Message 2 of 11 , Aug 1, 2006
      • 0 Attachment
        > Warning - Shameless plug for my book coming.


        Eric - for item number 1. Me and Mur cover what the splits are for
        many of the Automatic Ad Insertion networks, in our book Tricks of
        the Podcasting Masters.

        Chapter 18 starting around page 300.

        We cover about 10 of the networks.

        Rob W
        podCast411


        > _____
        >
        > From: podcasters@yahoogroups.com
        > [mailto:podcasters@yahoogroups.com] On
        > Behalf Of ericmattson1
        > Sent: Tuesday, August 01, 2006 6:05 AM
        > To: podcasters@yahoogroups.com
        > Subject: [podcasters] Dynamic Ad Insertion
        >
        > All,
        >
        > I've been doing some research into the companies offering dynamic ad
        > insertion and I have three questions I was hoping someone (podcaster
        > or vendor) could answer. (I tried searching the archives for the
        > answers but couldn't find them).
        >
        > 1. What are the % revenue splits of the major companies that do
        > dynamic ad insertion? Maybe I'm blind but the companies that do it
        > don't seem to want to disclose their splits publicly.
        >
        > 2. Is there a company that sells just the technology to do dynamic ad
        > insertion or do I have to go through one of the above vendors?
        >
        > 3. If one already has an advertiser, is it possible to negotiate a
        > lower rate (i.e. free money) with the suppliers of dynamic ad
        > insertion?
        >
        > I'd love any help or insight you might be able give me.
        >
        > Cheers,
        > Eric Mattson
        > Marketingmonger.com
        >
        > [Non-text portions of this message have been removed]
        >
        >
        >



        [Non-text portions of this message have been removed]
      • bryannpr
        ... The info was good, though! :) A few other additions: * Make sure the bitrates for the segments are the same, and ideally the normalization. Nothing more
        Message 3 of 11 , Aug 2, 2006
        • 0 Attachment
          --- In podcasters@yahoogroups.com, "Stephen Eley" <SFEley@...> wrote:
          >
          > Well, that was stupid.
          >
          > Oh well. If anyone wants my number or Skype handle, I guess you have
          > it now. No big deal. Good thing I didn't have a reason to e-mail him
          > my Social Security number!

          The info was good, though! :)

          A few other additions:

          * Make sure the bitrates for the segments are the same, and ideally the
          normalization. Nothing more jarring then soft content, LOUD AD, soft
          content.

          * Be careful to analyze the resulting file to make sure you have
          preserved any embedded ID3 info. At NPR we ran into an issue where our
          intro segment was encoded with LAME, which adds a specific header area
          that stores the duration of the file. MP3 cat does not re-write this
          field, so if your intro was 4 seconds and you stitch another 30 minutes
          of content, the header still says 4 seconds. One of the recent
          iterations of iTunes started paying attention to this field, and thus,
          all podcasts quit after 4 seconds. Not a good user experience...

          * Watch your meta. Stitching can wipe it out. And sometimes, things
          like embedded images will cause a stitch to fail. You may need to clean
          each segment, then stitch, then re-encode the meta.

          Bryan
        • Stephen Eley
          ... Strong agreement on the normalization. There are some tools that can help with this, like MP3Gain. (And I know Doug Kaye s team built a great
          Message 4 of 11 , Aug 2, 2006
          • 0 Attachment
            On 8/2/06, bryannpr <bryannpr@...> wrote:
            >
            > * Make sure the bitrates for the segments are the same, and ideally the
            > normalization. Nothing more jarring then soft content, LOUD AD, soft
            > content.

            Strong agreement on the normalization. There are some tools that can
            help with this, like MP3Gain. (And I know Doug Kaye's team built a
            great uploader/normalizer that I hope they'll open up to the public
            sometime.) >8->

            Bit rate -- it's a good idea to make sure you're consistent with the
            bit rate and sample rate, but I did some 'curiosity' testing on
            concatenating multiple MP3 files with different bit and sample rates
            together, and they played fine in the couple of players I tried.
            Combining two files with different bit rates turns your file into a
            VBR (variable bitrate) file, which almost all modern players know how
            to handle. Some of them might get confused about the duration,
            though, or express funny behavior when you pause or seek in the file.
            I don't know what you call it when you combine two files with
            different sample rates, except "silly."

            In any case, if you have control over your content it's a much better
            idea to keep all the encoding parameters the same. It's not strictly
            necessary, but it reduces a lot of risk of somebody's player breaking
            somewhere.


            > * Be careful to analyze the resulting file to make sure you have
            > preserved any embedded ID3 info.

            Very good point. This would be pretty easy to do with a dynamic
            script. (There are modules in Perl and many other languages for
            reading and writing ID3 tags.)


            > * Watch your meta. Stitching can wipe it out. And sometimes, things
            > like embedded images will cause a stitch to fail. You may need to clean
            > each segment, then stitch, then re-encode the meta.

            Really? I didn't know that! Thinking about it, I suppose a really
            *large* amount of metadata could overload a player's frame buffer and
            convince it there weren't any more sound frames. The problem with
            making assumptions about this would be the testing required on every
            player... So yes, I think you're right. The safest course is just to
            programmatically take all tags out and put the "right" ones back at
            the beginning.

            Thanks for the practical insight! I really enjoy geeking on the
            details of things like this...


            --
            Have Fun,
            Steve Eley (sfeley@...)
            ESCAPE POD - The Science Fiction Podcast Magazine
            http://www.escapepod.org
          • Michael Elsdoerfer
            ... Basically, yes, that s it. In real life it becomes slightly more complicated, though (as is probably always the case ;)). Bryan mentioned most of it,
            Message 5 of 11 , Aug 2, 2006
            • 0 Attachment
              > 'cat FirstHalf.mp3 AdToInsert.mp3 SecondHalf.mp3 > FinalFile.mp3'

              Basically, yes, that's it. In real life it becomes slightly more
              complicated, though (as is probably always the case ;)).

              Bryan mentioned most of it, frequency, bitrate, id3 tags (v1 and v2 might
              require different handling, depending on how you do it), but also HTTP
              headers, caching, download resuming etc. need to be considered.

              Doing all that on every request is really inefficient, too. Be sure to at
              least cache each combination of ad and episode, or even better, forget cat &
              friends and create a module for the webserver of your choice. It's
              surprisingly simple to write an Apache output filter to accomplish "real"
              on-the-fly mp3 merging, without relevant performance penalties.

              In fact, the guy who, as I understand it, wrote the module Kiptronic is
              offering/using has published a couple of articles on the subject on ONLamp.

              Michael
            • Adam Varga
              We ve got a PHP script for stitching MP3s http://www.dailysonic.com/manglepod/ Feel free to use it in terms for whatever... its a GPL License, meaning you can
              Message 6 of 11 , Aug 6, 2006
              • 0 Attachment
                We've got a PHP script for stitching MP3s
                http://www.dailysonic.com/manglepod/
                Feel free to use it in terms for whatever... its a GPL License,
                meaning you can use it for commercial uses too... modify it....

                Also... in terms of splits...
                You might wanna check out this comparison chart
                http://www.okaytoplay.com/podcast-advertising-companies/

                - Adam

                On 8/2/06, Michael Elsdoerfer <elsdoerfer@...> wrote:
                >
                >
                >
                >
                >
                >
                > > 'cat FirstHalf.mp3 AdToInsert.mp3 SecondHalf.mp3 > FinalFile.mp3'
                >
                > Basically, yes, that's it. In real life it becomes slightly more
                > complicated, though (as is probably always the case ;)).
                >
                > Bryan mentioned most of it, frequency, bitrate, id3 tags (v1 and v2 might
                > require different handling, depending on how you do it), but also HTTP
                > headers, caching, download resuming etc. need to be considered.
                >
                > Doing all that on every request is really inefficient, too. Be sure to at
                > least cache each combination of ad and episode, or even better, forget cat
                > &
                > friends and create a module for the webserver of your choice. It's
                > surprisingly simple to write an Apache output filter to accomplish "real"
                > on-the-fly mp3 merging, without relevant performance penalties.
                >
                > In fact, the guy who, as I understand it, wrote the module Kiptronic is
                > offering/using has published a couple of articles on the subject on ONLamp.
                >
                > Michael
                >
                >
                >
                >


                --
                DAILYSONIC feed your ears
                http://www.dailysonic.com
                877 SONIC LINE
              Your message has been successfully submitted and would be delivered to recipients shortly.