The WeaveSignal contributors
16833c96ae
- tests/demo-video/: the script, designed voices, the run through the real desk and the edit that make the website's demo video (the media themselves are not in git). - The home page plays the video, with its credits. - deploy/cloudron/CHANGELOG.md: one line per entry; CloudronVersions.json rebuilt.
26 lines
1.3 KiB
Shell
26 lines
1.3 KiB
Shell
# SPDX-License-Identifier: AGPL-3.0-only
|
||
# SPDX-FileCopyrightText: 2026 The WeaveSignal contributors
|
||
#
|
||
# Comparing what Whisper heard with the script, for synth.sh and check.sh (sourced; needs
|
||
# $script, the path of script.json).
|
||
|
||
# Words for comparing: lower case, no punctuation, each name's spellings folded to one.
|
||
norm() {
|
||
jq -rn --arg t "$1" --arg q "'" --slurpfile s "$script" '
|
||
($s[0].heard_as | to_entries) as $names
|
||
| $t | ascii_downcase | gsub("[‘’" + $q + "]"; "") | gsub("-"; " ")
|
||
| gsub("[^a-z0-9 ]"; " ")
|
||
| reduce $names[] as $n (.; reduce ($n.value[] | gsub("-"; " ") | gsub($q; "")) as $alt (.; gsub("\\b" + $alt + "\\b"; ($n.key | ascii_downcase | gsub("-"; " ")))))
|
||
| gsub("nineteen fifty six"; "1956") | gsub("\\b(ten|10) past eleven\\b"; "1110")
|
||
| [splits(" +")] | map(select(length > 0)) | join(" ")'
|
||
}
|
||
# Word edit distance between two normalised strings.
|
||
distance() {
|
||
jq -rn --arg a "$1" --arg b "$2" '
|
||
($a | split(" ")) as $x | ($b | split(" ")) as $y
|
||
| reduce range(0; $x | length) as $i ([range(0; ($y | length) + 1)];
|
||
. as $prev
|
||
| reduce range(0; $y | length) as $j ([$i + 1];
|
||
. + [[$prev[$j + 1] + 1, .[$j] + 1, $prev[$j] + (if $x[$i] == $y[$j] then 0 else 1 end)] | min]))
|
||
| .[-1]'
|
||
}
|