[opus] Opus for ASR
Milan.Young at nuance.com
Mon Sep 17 12:12:29 PDT 2012
> From: Jean-Marc Valin [mailto:jmvalin at jmvalin.ca]
> On 12-09-14 07:15 PM, Benjamin Schwartz wrote:
> >> One could imagine eventually rolling these updates back into the
> >> standard under an "ASR" mode.
> > This seems very unlikely to me. Opus is a decoder-specified standard,
> > so the encoder can be modified arbitrarily without requiring
> > re-standardization. It's hard to imagine anything worth doing that
> > would cause you to go outside the current standard.
> What Milan (presumably) meant was adding these to the opus-codec.org
> implementation, not the IETF RFC.
[Milan] In the short term, we'll probably be looking to replace our Speex encoder with an Opus version. Since we'll be controlling the encoding process we would be free to select from the upcoming flags exposed by opus-codec.org. This would probably give us a small (but still welcome) boost in accuracy relative to Speex.
But another significant selling point of Opus is its availability in standards-based platforms like web browsers. And although we can probably expect most platforms to adopt the implementation shared by opus-codec.org, we would need a mechanism for the application to specify that it wants the ASR friendly encoding. It might be able to pass that as a flag on the MIME type, but without a standard to describe those flags, I'm pessimistic that browsers will support it.
Is there any precedent for standard flags associated with a codec ( eg bit rate)? Any other good reasons for including aspects of encoding in 6716 or pushing through a sibling RFC?
More information about the opus