eric/seriatim
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 4 Wiki Activity

Added a module to coalesce adjacent same-speaker segments

Browse Source
This commit is contained in:
Eric Rakestraw
2026-04-27 19:30:00 -05:00
parent 13d972cb24
commit aab6d12730
12 changed files with 919 additions and 28 deletions
Download Patch File Download Diff File Expand all files Collapse all files

116
internal/overlap/resolve_test.go
View File

@@ -2,6 +2,7 @@ package overlap
import (
"reflect"
"strings"
"testing"
"gitea.maximumdirect.net/eric/seriatim/internal/model"
@@ -14,7 +15,7 @@ func TestResolveNoOverlapGroupsIsNoOp(t *testing.T) {
},
}
got, summary, err := Resolve(merged, 0.75)
got, summary, err := Resolve(merged, 0.75, 0.4)
if err != nil {
t.Fatalf("resolve failed: %v", err)
}
@@ -39,7 +40,7 @@ func TestResolveCreatesChronologicalWordRunSegments(t *testing.T) {
merged.Segments[0].OverlapGroupID = 1
merged.Segments[1].OverlapGroupID = 1
got, summary, err := Resolve(merged, 0.75)
got, summary, err := Resolve(merged, 0.75, 0.4)
if err != nil {
t.Fatalf("resolve failed: %v", err)
}
@@ -93,7 +94,7 @@ func TestResolveIncludesWordsByIntervalIntersection(t *testing.T) {
}
merged.Segments[0].OverlapGroupID = 1
got, _, err := Resolve(merged, 10)
got, _, err := Resolve(merged, 10, 0.4)
if err != nil {
t.Fatalf("resolve failed: %v", err)
}
@@ -116,7 +117,7 @@ func TestResolveWordRunGapThreshold(t *testing.T) {
}
merged.Segments[0].OverlapGroupID = 1
got, _, err := Resolve(merged, 0.75)
got, _, err := Resolve(merged, 0.75, 0.4)
if err != nil {
t.Fatalf("resolve failed: %v", err)
}
@@ -141,7 +142,7 @@ func TestResolvePartialResolutionKeepsNoWordSpeakerOriginals(t *testing.T) {
merged.Segments[0].OverlapGroupID = 1
merged.Segments[1].OverlapGroupID = 1
got, summary, err := Resolve(merged, 0.75)
got, summary, err := Resolve(merged, 0.75, 0.4)
if err != nil {
t.Fatalf("resolve failed: %v", err)
}
@@ -178,7 +179,7 @@ func TestResolveGroupWithNoUsableWordsRemainsUnchanged(t *testing.T) {
merged.Segments[0].OverlapGroupID = 1
merged.Segments[1].OverlapGroupID = 1
got, summary, err := Resolve(merged, 0.75)
got, summary, err := Resolve(merged, 0.75, 0.4)
if err != nil {
t.Fatalf("resolve failed: %v", err)
}
@@ -201,7 +202,7 @@ func TestResolveReplacementProvenanceIsDeterministic(t *testing.T) {
},
}
got, _, err := Resolve(merged, 0.75)
got, _, err := Resolve(merged, 0.75, 0.4)
if err != nil {
t.Fatalf("resolve failed: %v", err)
}
@@ -238,7 +239,7 @@ func TestResolveIncludesUntimedWordsInTextWithoutChangingBounds(t *testing.T) {
},
}
got, _, err := Resolve(merged, 0.75)
got, _, err := Resolve(merged, 0.75, 0.4)
if err != nil {
t.Fatalf("resolve failed: %v", err)
}
@@ -273,7 +274,7 @@ func TestResolveUntimedWordsDoNotBridgeWordRunGap(t *testing.T) {
},
}
got, _, err := Resolve(merged, 0.75)
got, _, err := Resolve(merged, 0.75, 0.4)
if err != nil {
t.Fatalf("resolve failed: %v", err)
}
@@ -299,7 +300,7 @@ func TestResolveSpeakerWithOnlyUntimedWordsIsNotReplaced(t *testing.T) {
}
merged.Segments[0].OverlapGroupID = 1
got, summary, err := Resolve(merged, 0.75)
got, summary, err := Resolve(merged, 0.75, 0.4)
if err != nil {
t.Fatalf("resolve failed: %v", err)
}
@@ -311,6 +312,93 @@ func TestResolveSpeakerWithOnlyUntimedWordsIsNotReplaced(t *testing.T) {
}
}
func TestResolveReordersNearStartWordRunsByDuration(t *testing.T) {
merged := model.MergedTranscript{
Segments: []model.Segment{
segmentWithWords("a.json", 0, "Alice", 1, 3, word("long", 1, 2)),
segmentWithWords("b.json", 0, "Bob", 1, 3, word("short", 1.2, 1.3)),
},
OverlapGroups: []model.OverlapGroup{
group(1, 1, 3, []string{"a.json#0", "b.json#0"}, []string{"Alice", "Bob"}),
},
}
got, _, err := Resolve(merged, 0.75, 0.4)
if err != nil {
t.Fatalf("resolve failed: %v", err)
}
if gotTexts(got.Segments) != "short,long" {
t.Fatalf("segment order = %s, want short,long", gotTexts(got.Segments))
}
if got.Segments[0].Start != 1.2 || got.Segments[0].End != 1.3 {
t.Fatalf("short segment bounds changed: %#v", got.Segments[0])
}
if got.Segments[1].SourceRef != "word-run:1:1:1" || got.Segments[1].Text != "long" {
t.Fatalf("long segment provenance/text changed: %#v", got.Segments[1])
}
}
func TestResolveDoesNotReorderWordRunsOutsideWindow(t *testing.T) {
merged := model.MergedTranscript{
Segments: []model.Segment{
segmentWithWords("a.json", 0, "Alice", 1, 3, word("long", 1, 2)),
segmentWithWords("b.json", 0, "Bob", 1, 3, word("short", 1.5, 1.6)),
},
OverlapGroups: []model.OverlapGroup{
group(1, 1, 3, []string{"a.json#0", "b.json#0"}, []string{"Alice", "Bob"}),
},
}
got, _, err := Resolve(merged, 0.75, 0.4)
if err != nil {
t.Fatalf("resolve failed: %v", err)
}
if gotTexts(got.Segments) != "long,short" {
t.Fatalf("segment order = %s, want long,short", gotTexts(got.Segments))
}
}
func TestResolveReordersTransitiveNearStartClustersByDuration(t *testing.T) {
merged := model.MergedTranscript{
Segments: []model.Segment{
segmentWithWords("a.json", 0, "Alice", 1, 3, word("long", 1, 2)),
segmentWithWords("b.json", 0, "Bob", 1, 3, word("medium", 1.3, 1.8)),
segmentWithWords("c.json", 0, "Carol", 1, 3, word("short", 1.65, 1.75)),
},
OverlapGroups: []model.OverlapGroup{
group(1, 1, 3, []string{"a.json#0", "b.json#0", "c.json#0"}, []string{"Alice", "Bob", "Carol"}),
},
}
got, _, err := Resolve(merged, 0.75, 0.4)
if err != nil {
t.Fatalf("resolve failed: %v", err)
}
if gotTexts(got.Segments) != "short,medium,long" {
t.Fatalf("segment order = %s, want short,medium,long", gotTexts(got.Segments))
}
}
func TestResolveReorderFallsBackToDeterministicOrderForEqualDurations(t *testing.T) {
merged := model.MergedTranscript{
Segments: []model.Segment{
segmentWithWords("b.json", 0, "Bob", 1, 3, word("bob", 1, 1.5)),
segmentWithWords("a.json", 0, "Alice", 1, 3, word("alice", 1.2, 1.7)),
},
OverlapGroups: []model.OverlapGroup{
group(1, 1, 3, []string{"b.json#0", "a.json#0"}, []string{"Bob", "Alice"}),
},
}
got, _, err := Resolve(merged, 0.75, 0.4)
if err != nil {
t.Fatalf("resolve failed: %v", err)
}
if gotTexts(got.Segments) != "bob,alice" {
t.Fatalf("segment order = %s, want bob,alice", gotTexts(got.Segments))
}
}
func segmentWithWords(source string, sourceIndex int, speaker string, start float64, end float64, words ...model.Word) model.Segment {
segment := segment(source, sourceIndex, speaker, start, end)
segment.Words = words
@@ -326,6 +414,14 @@ func word(text string, start float64, end float64) model.Word {
}
}
func gotTexts(segments []model.Segment) string {
texts := make([]string, 0, len(segments))
for _, segment := range segments {
texts = append(texts, segment.Text)
}
return strings.Join(texts, ",")
}
func untimedWord(text string) model.Word {
return model.Word{
Text: text,