Non-proprietary audio formats for CAP profile

["Art Botterell" <ABott@so.cccounty.us>]

As you're aware, one of the comments we've received on the draft profile has to do with the audio encoding format or formats supported for recorded and streaming audio, primarily for EAS.  This is important because whatever format or formats are specified in the profile will, presumably, become mandatory for any next-generation EAS implementation.  

The ECIG (a semi-formal working group of some the major EAS equipment manufacturers) recommended two audio formats: WAV and MP3.  The WAV format is non-proprietary but requires much more bandwidth than compressed formats such as MP3.  The MP3 format is more bandwidth-efficient than WAV, but it's subject to patent restrictions and would require implementers to enter into licensing agreements and pay fees.

Commenting on the Profile draft (# 69 in our current issues list), Tom Wood of Digital Alert Systems, an EAS manufacturer and ECIG member, suggested that the non-proprietary Ogg/Vorbis encoding be added to the list of acceptable formats.  He argued that this format can provide audio quality and data compression comparable to MP3 but without patent restrictions.  He also suggested the addition of the Ogg/Theora format for video.

A note on the nomenclature:  "Ogg" refers to a file format that can be used to store multimedia content in a number of encodings.  "Vorbis," "Theora" and also "Speex" (see below) refer to encoding techniques for encoding analog content such as audio or video in digital form.  In the case of fixed-length recordings an encoded bitstream can be packaged in an Ogg-format file; thus, for example, the label "ogg/vorbis."  However, in a streaming-media application the encoding frequently is transmitted without the Ogg encapsulation, e.g., in an RTP-based streaming system such as IceCast or ShoutCast.

In reviewing the literature on the Vorbis encoding it came to my attention that there is another codec in the same series of open-source, non-proprietary formats, one known as "Speex," that is more highly optimized for voice transmission over limited bandwidth links (2-44 kbps).  The Speex webside explains that "Vorbis is mostly aimed at compressing music and audio in general, which Speex targets speech only.  For that reason Speex can achieve much better results than Vorbis on speech, typically 2-4 times higher compression at equal quality."  A few links for reference:

* The Speex website:		<http://speex.org >

* Some applications of Speex:	<http://en.wikipedia.org/wiki/Speex#Applications >

* The Speex open-source license:	<http://speex.org/docs/manual/speex-manual/node15.html#sec:Speex-License>

As it happens I've done a bit of experimenting recently with a Speex implementation (TeamSpeak) for encoding the various audio from a high-grade radio scanner.  I've been impressed with the quality of a wide variety of source audio... male and female voices, noisy signals and full-quieting ones.  I'd describe the audio quality of a 17kbps Speex-encoded stream as being comparable with commercial AM broadcast, at least on speech signals.

So I think we have several questions to address:

1)  Should we support multiple encodings for audio, or standardize on one?  [Personally I'm not sure there's any reason to require implementers to support duplicates if we can resolve what I think was being viewed as an unavoidable trade-off between efficiency and proprietary encoding.]

2)  Should we specify a video encoding at this time?  [Since the Xipf series of codecs offers a mature non-proprietary ogg-compatible format, called Theora, I'd be inclined to specify it in order to encourage the development of video in warning.]

3)  What preference, if any, should we give to non-proprietary encodings? [I'd suggest that the use of patent-free formats is appropriate to encourage wide adoption of the Profile, particularly given that our recommendation will have quasi-regulatory effect in the U.S.]

I look forward to discussing this on our call tomorrow.

- Art



Art Botterell, Manager
Community Warning System
Contra Costa County Office of the Sheriff
50 Glacier Drive
Martinez, California 94553
(925) 313-9603
fax (925) 646-1120