Skip to content

keyword detection: explain formula for keyphrase buffer size - #182

Merged
lgirdwood merged 1 commit into
thesofproject:masterfrom
lbetlej:kpd-latency
Mar 2, 2020
Merged

lgirdwood merged 1 commit into
thesofproject:masterfrom
lbetlej:kpd-latency

Conversation

@lbetlej

@lbetlej lbetlej commented Feb 5, 2020

Copy link
Copy Markdown
Contributor

An explanation of various types of latencies that determine minimum size of
a keyphrase buffer is added. It includes information about differences in
movement between write and read pointers to the buffer.

Signed-off-by: Lech Betlej lech.betlej@linux.intel.com

Write Pointer (WP) is used to indicate where data coming from microphones is
written to a keyphrase buffer. The keyphrase buffer is organized as
a cyclic buffer and WP moves as long as data is coming from mics at
a regular rate. Read pointer (RP) indicates from which offset in teh buffer

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

typo, s/teh/the

a regular rate. Read pointer (RP) indicates from which offset in teh buffer
data is fetched to host. To start burst transmission RP is set to WP
position. RP moves faster then WP as draining is executed as a background
task until RP reaches again WP that moves a a regular (slower) rate. This

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

typo, s/a/at

@mrajwa mrajwa Feb 5, 2020 •

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Also, here we claim that we start draining from WP (write pointer) which is not true - in fact we calculate read position base on host client algorithm / topology param called "history depth".

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@lbetlej Please let me know when/if you make this update based on @mmaka1 's remarks. Thanks

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@lbetlej @lgirdwood Please let me know if this comment by @mrajwa has been resolved so I can merge this PR. [...in fact we calculate read position base on host client algorithm / topology param called "history depth"]

Thanks.

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@mrajwa please respond.

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@lgirdwood there is nothing more I can say/do in regards to this PR. We have to wait for an update from @lbetlej. He is on sick leave now so let's give him few days until his well.

@deb-intel deb-intel Feb 12, 2020 •

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@mrajwa Thanks for your update. In the interest of time, I am going to merge PRs 182-183 and also add my own edits. I will make a note of @mrajwa's comments. @lbetlej can update the new page when he returns. Thanks, all.

UPDATE: This branch has conflicts; I suspect they are from PR #183 so I am going to hold off on editing kd-integration.rst until after @lbetlej returns.

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@deb-intel I've rebased and updated the PR to address all findings listed in the conversation. Could you wait for review approval from @mrajwa before you start proof reading ?

@lbetlej lbetlej changed the title keyword detection: explain formula for keyphrase buffer size [WIP] keyword detection: explain formula for keyphrase buffer size Feb 6, 2020

@lgirdwood lgirdwood left a comment

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Lets get the changes proposed by @mrajwa fixed and we are good to merge.

@lbetlej lbetlej changed the title [WIP] keyword detection: explain formula for keyphrase buffer size keyword detection: explain formula for keyphrase buffer size Feb 28, 2020
3. L3 is a time required to send already buffered data to host. Typically
Write Pointer (WP) is used to indicate where data coming from microphones is
written to a keyphrase buffer. The keyphrase buffer is organized as
a cyclic buffer and WP move if data is coming from mics at

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

s/move/moves

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

ack

@mrajwa

mrajwa commented Feb 28, 2020

Copy link
Copy Markdown
Contributor

Functional wise its fine for me now.

An explanation of various types of latencies that determine minimum size of
a keyphrase buffer is added. It includes information about differences in
movement between write and read pointers to the buffer.

Signed-off-by: Lech Betlej <lech.betlej@linux.intel.com>
@lgirdwood
lgirdwood merged commit 5d32e4c into thesofproject:master Mar 2, 2020
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants