Find millimeter on Facebook

Related Articles

Encoding for podcasts

Dec 21, 2010 12:27 PM, By Jan Ozer


      Subscribe in NewsGator Online   Subscribe in Bloglines  

A few editions ago, we discussed choosing a resolution for podcasts delivered to iDevices via iTunes. I started by reviewing the specifications for the various iDevices, and then working from the results of about 50 files that I had downloaded from iTunes, I related the resolution-related strategies used by various iTunes producers, from Dr. Ruth to Oprah.

That story was intended to help you choose a resolution (or resolutions) for your podcasts. In this story, I’ll discuss the encoding parameters used by those producers and compare them to presets offered in Apple Compressor, Sorenson Squeeze version 6.5 and Telestream Epsiode Pro version 6.0.2.

To briefly recount my methodology, I downloaded a total of 53 free podcasts that I classified in four basic groups; 320x240/320x180 or smaller, 320x240/320x180 to 480x270, 640x360 to 640x480, and podcasts larger than 640x480. As you’ll see in the tables below, I analyzed the files to determine configuration options like average data rate, H.264 profile, key frame setting and the like, to learn how real producers are making their files available on iTunes. Then I compared them to presets from the aforementioned three products.

None of the encoders offered presets for the 480x270 resolution, which probably reflects that this resolution, originally developed for the first-generation iPhone and iPod Touch devices, is no longer that widely used. Instead, most producers encode at 640x360 resolution, producing a file that is still compatible with these older devices, but will look better when played on devices with larger resolution displays, like iPads, iPhone/iTouch 4G and computers. In this category, I’ll share the configuration options used in files that I downloaded, but have no other data to share.

320x240/320x180 files
In this category, three of 10 producers rendered at 320x180, with the bulk preferring the 4:3, 320x240 resolution. As you can see in the templates, the Baseline profile is mandatory because most older iPods are spec-ed at Baseline, and files encoded using the Main profile likely won’t load on these older devices. CABAC isn’t available when encoding with the Baseline profile, leaving CAVLC entropy encoding as the only option.

Survey

Apple Compressor

Sorenson Squeeze

Telestream
Episode

Preset name

Not applicable

H.264 for iPod/iPhone - 320x240

iPod_Low

iPod 320x240

Profile

Baseline (8)
Main (1)

Baseline

Baseline

Baseline

Entropy

CAVLC

CAVLC

CAVLC

CAVLC

Data rate - target

534kb/s

600kb/s

600kb/s

600kb/s

Data rate - max

Not available

768kb/s

636kb/s

1440kb/s

Frame rate

29.97 (9)
15 (1)

Same as source

Same as source

29.97

Key frame Interval (frames)

150

150

300

200

Audio data rate

113kb/s

128kb/s

128kb/s

128kb/s

Channels

Stereo

Stereo

Stereo

Stereo


Table 1: 320x240 results

All encoders used variable bit rate encoding for iDevices, specifying both a target and maximum data rate. Unfortunately, my file analysis tool — Inlet Semaphore for those who care about such things — could only tell me the average bitrate for the survey files, not that maximum, which is why that field is “Not available” in the survey column.

The target data rate of 600kb/s for the three encoders is reasonably close to the 534kb/s average found in the 10 files in this category, which varied from a low of 256k/ps and high of 846kb/s. If you’re encoding with VBR, set the maximum data rate to 768kb/s, which matches the Level 1.3 maximum of the oldest iPods; otherwise, your files may not play on those devices. You should dial that down manually in Episode.

Eight of the 10 files analyzed were produced at full frame rate, with one at 15fps. I recommend using the full frame rate, as do the presets. Though it looks like Episode will convert all source fies to 29.97 files, in my tests, it didn’t change the frame rate of the 24p file that I input, so that shouldn’t be a problem.

Probably the biggest mistake that I found in the surveyed files was the key frame interval, which was 30 frames or lower in 10 of the files. Key frames, which are otherwise known as I-frames, are the least efficient frame of the H.264 IBP frame types, and the key frame interval should be set in the 150 - 300 range shown in the templates. Finally, the actual audio bit rate of 113kb/s matched the presets well, as did the stereo designation.

Share this article




Continue the discussion on Crosstalk the Millimeter Forum.


© 2014 NewBay Media, LLC.

Browse Back Issues
Back to Top