Issue metadata
Sign in to add a comment
|
Move high resolution speech files from rootfs to stateful partition |
||||||||||||||||||||||
Issue descriptionCurrently the files in src/platform/assets/speech_synthesis/patts are all high resolution voice files for a11y features of Chrome OS, But they constitute of about 120MB of zipped files. These files are rarely used and cannot be compressed too much anymore. The suggestion is to move these files into stateful partition from rootfs. This will reduce the size of the rootfs partition by around 120MB and reduces the size of full payloads by around 100 MB which is significant. Moving these files into stateful partition could mean that these files can be downloaded whenever they are accessed first. But, we need some form of text to speech files be present for a visually impaired person the first time they open the chromebook. This could be either a much smaller set of low quality files be present in rootfs and high quality files be downloaded later or the high quality files be always present in the stateful partition out of factory and on recoveries. Component updaters can be used to handled download/upgrade of these files. This is just a feature request to track this issue.
,
Nov 8 2017
,
Nov 9 2017
,
Jan 17 2018
Just revisiting this issue. A question for the UI>Acceessibility team, how much work does this feature needs? componentizing this speech files is possible for us now and this is a very good example of a chrome os component. Is there any way we can move this forward? Can we (component updater team) help with this feature?
,
Jan 17 2018
Adding katie@ and dtseng@ The blocker here is that we need something that's part of rootfs. I've started a port of espeak (a small, free, low-quality speech engine). Once that's done we should be able to move the speech engine to the stateful partition.
,
Jan 17 2018
Great news, Good to know that. Please keep us updated on this :) Thank you.
,
Sep 21
I don't believe we would want to move the higher quality files to stateful. The usage is perhaps small, but the impact is crucial as without text to speech, there's no way these users can access the device. With that said, we recently did remove some voice data files and implement downloadable voices within Chrome settings, so the current size will never increase. This would be a significant regression in the user experience for anyone who initially starts ChromeVox or any of the other services that want to use text to speech.
,
Sep 21
I think the plan is to use DLC for this purpose to cover out of factory experiences. DLCs would allow support for many more languages.
,
Sep 28
|
|||||||||||||||||||||||
►
Sign in to add a comment |
|||||||||||||||||||||||
Comment 1 by xiaochu@chromium.org
, Nov 8 2017