add backchannel annotation file

phiresky · Jan 22, 2021 · 6e6e671 · 6e6e671
1 parent 4668fe5
commit 6e6e671
Show file tree

Hide file tree

Showing 4 changed files with 487,369 additions and 0 deletions.
diff --git a/README.md b/README.md
@@ -22,8 +22,11 @@ Then open <http://localhost:3000> in the browser (loading takes a bit).
 
 Includes the live demo (microphone input!) and some sample tracks from the Switchboard data set: sw2249, sw2258, sw2411, sw2485, sw273,  sw2807, w2254, sw2297, sw2432, sw2606, sw2762, sw4193. Selecting other tracks will fail.
 
+## Data
 
 
+The data/utterance_is_backchannel.json contains a dump of backchannel vs non-backchannel data for all utterances in Switchboard, generated by `python -m extract.standalone_save_backchannels_to_json`.
+
 ## Evaluation Visualizer
 
 ### Screenshots: 

diff --git a/data/README.md b/data/README.md
@@ -17,3 +17,5 @@ The original transcriptions are from <https://www.isip.piconepress.com/projects/
 You can compare the integrity of the data files with the ones I used by running
 
     sha256sum -c sha256sums.txt
+
+The utterance_is_backchannel.json contains a dump of backchannel vs non-backchannel data for all utterances in Switchboard, generated by `python -m extract.standalone_save_backchannels_to_json`.
Original file line number	Diff line number	Diff line change
Expand Up		@@ -17,3 +17,5 @@ The original transcriptions are from <https://www.isip.piconepress.com/projects/
		You can compare the integrity of the data files with the ones I used by running

		sha256sum -c sha256sums.txt

		The utterance_is_backchannel.json contains a dump of backchannel vs non-backchannel data for all utterances in Switchboard, generated by `python -m extract.standalone_save_backchannels_to_json`.