ABCDEFGHIJKLMNOPQRSTUVWXYZ
1
filedata release flavordescription
2
cb18ac98-a58c-4599-8bc8-461dab667322--The folder name is a random unique ID used to identify a specific conversation.
3
├── audio_video_features.csvno_mediaAcoustic and visual features calculated at a 1-second frequency throughout the conversation.
4
├── metadata.jsonno_mediaMetadata for the conversation including speaker IDs, audio channels, and raw recording information.
5
├── processedprocessed_mediaFolder containing processed video files, audio files, and metadata.
6
│ ├── 57272517e386b9000e3e2b72.mp4processed_mediaProcessed video for a single speaker (file name is <user_id>.mp4).
7
│ ├── 5cb60f1eefd4130001bbfdb7.mp4processed_mediaProcessed video for the other speaker (file name is <user_id>.mp4).
8
│ ├── cb18ac98-a58c-4599-8bc8-461dab667322.mp3processed_mediaProcessed audio file of the conversation containing both speakers, one in the left channel and the other in the right channel (see channel_map.json). File name is <conversation_id>.mp3.
9
│ ├── cb18ac98-a58c-4599-8bc8-461dab667322.mp4processed_mediaProcessed video file of the conversation containing both speakers side-by-side; audio for one in the left channel and the other in the right channel (see channel_map.json). File name is <conversation_id>.mp4.
10
│ ├── channel_map.jsonprocessed_mediaMetadata that maps a specific user id to the left and right audio channels for the processed video and audio files, and the side of the frame the speaker is on for processed video where the speakers appear side-by-side.
11
│ └── thumbnail.pngprocessed_mediaAn image thumbnail of the processed video.
12
├── rawraw_mediaFolder containing the raw recordings of the conversations (separate video files for each speaker); may be more video files if speaker disconnected and rejoined the conversation.
13
│ ├── 1c5b8b8e-4bf3-47b7-bb02-b20132edb41e.mkvraw_mediaRaw recording stream file of a single speaker. Associated metadata for user id, and start/stop times is in raw/<conversation_id.json>. File name is <stream_id.mkv>.
14
│ ├── 97ee4cbb-b70b-4fb4-8d64-a468efc560e0.mkvraw_media
Raw recording stream file of a single speaker. If there are two files, this is the other speaker. If more than two, it could be either speaker. Associated metadata for user id, and start/stop times is in raw/<conversation_id.json>. File name is <stream_id.mkv>.
15
│ └── cb18ac98-a58c-4599-8bc8-461dab667322.jsonraw_mediaRaw metadata for each of the stream files in the raw folder.
16
├── survey.csvno_mediaAnswers to pre- and post-conversation survey questions by the participants in the conversation (if available).
17
├── transcriptionno_mediaTranscription of the conversation.
18
│ ├── transcribe_output.jsonno_mediaThe raw JSON output returned from the AWS Transcribe service for this conversation; contains token-level timings and confidences.
19
│ ├── transcript_audiophile.csvno_mediaTranscript of the conversation compiled from transcribe_output.json into a CSV with a row for each turn in the conversation.
20
│ ├── transcript_backbiter.csvno_mediaTranscript of the conversation using the 'backbiter' method as described in the paper.
21
│ └── transcript_cliffhanger.csvno_mediaTranscript of the conversation using the 'cliffhanger' method as described in the paper.
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100