Reviewed: Adobe Creative Suite 4 Production Premium
Having chosen H.264 Blu-ray and Dolby digital audio for our compression, we let Media Encoder load our timeline into its queue. Two hours later, our 3-minute test compression was done. In place of FLV and F4V files, which were having a few issues in our prerelease version, we exported a QuickTime streaming file at the full 1440x1080 size, which turned out to be excellent quality
Adobe Media Encoder can also upload a completed compression to a user-identified server, a significant step in workflow enhancement. However, as Tim mentioned in an article on StreamingMedia.com, automated FTP transmissions are only available on Adobe’s more expensive Flash Media Server. There is also the option to have Media Encoder delete the original file after verifying a successful upload to a remote location
Transcription With Premiere Pro and Soundbooth
Adobe has added an autotranscription feature to Premiere Pro and Soundbooth. In theory, you can take a video or audio clip and run it through the analyzer, which will spit out a speech-to-text metadata file that times each word to a synchronous frame of video. There are several tools that do this and do it quite well, but they range from a low-end price tag of about $18,000 to more than $100,000. Adobe licensed a set of speech-to-text libraries that are acceptable for the price range, even though they seem to lag behind other speech indexing, search, and retrieval systems we’ve used.
Adobe recommends using clean audio with minimal extraneous noise, meaning that you may want to use Soundbooth to clean up the audio before attempting the speech-to-text transcription. The program also has the ability to distinguish between more than one speaker and label each individually
In our tests, the autotranscription feature performed very poorly. It had 95%-plus accuracy when our speaker had noticeable breaks between words, but it dropped to the 50% range when he spoke at a natural rate with few pauses. While we could suggest it was the Southern accent, this wouldn’t account for the lack of accuracy, since the South is not exactly known for fast talkers. The transcript did make for some very humorous reading, though. We tried running a few other videos through the transcriber in hopes that they would have easier-to-transcribe audio tracks, but we didn’t get any better results
The transcript is also available as only metadata, with each word tied to a particular frame of video. So editing the transcript requires clicking on and changing every single word individually. It would be nice to see a modification that allows users to export the transcription to a text document for ease of distribution and editing, as some of the other systems on the market do, and to reimport the edited file with autoparsing (compared against the original transcription).
We like the transcription feature in concept, but we think Adobe can do better, especially with some of the systems out there having more than a decade of trial-and-error testing. Hopefully, these shortcomings will be improved as technology advances and future Adobe releases come along.
The Case for Metadata
If there’s any common thread that you’ve noticed in all of the CS4 programs thus far, it’s metadata. The dictionary definition of metadata is "a set of data that describes and gives information about other data." Not so descriptive, especially since we were always taught never to define a word with the same word. The practical benefits of Adobe tying metadata in at every level can’t be overstated, though, so let’s look at why Adobe has incorporated metadata entry and retrieval extensively into the new product release.
Probably the most common and well-known metadata today is the infamous ID3 tags that all of our digital music files come with. These wonderful little bits of information help us keep our iPods and media player libraries neat and tidy by providing all the track and album information for our music. But this same technology is available in any computer file. Look at the properties of any file on your hard drive, and you’ll see all kinds of information that you may not have known was there, such as the software developer’s name, the version number, a URL link to the company’s homepage, and so on
Adobe understands that having the power to edit this information about your original digital content can be extremely advantageous. The metadata that you add to your videos can, in theory, be carried into and displayed in every major media player that will be used to view your content. Not only is this metadata addiction a good marketing strategy, it also helps make the Adobe Media Player and interconnecting technologies such as Adobe’s Bridge tool more dynamic and easier to search. If you’re having trouble finding the video of the recent vice presidential candidates’ debate that your intern logged and tagged for you, try searching for keywords such as financial bailout, foreign policy, or doggone it, and you’ll understand that we’re entering an era where search is becoming a major part of workflow acceleration.
Impressive but Imperfect
In summary, although the release of CS4 pushes Adobe’s video software further into the spotlight, it doesn’t make as large a first impression as its predecessor did. Adobe Media Encoder, OnLocation, and Soundbooth in particular have seen some real growth, but Premiere Pro’s updates are significantly more subdued unless tapeless workflows (such as native AVCHD or the RED camera) are your primary method of acquisition
The speech-to-text and metadata tweaks make the upgrade especially attractive to content creators who do a significant amount of repurposing. More specifically, content creators who deal primarily in streaming video and who want to ensure that their content always gets distributed with credits and copyright information intact will be able to capitalize on metadata tagging for search engine optimization now that Flash files are searchable by major search engines such as Yahoo! and Google.