Initial push to gitea

internal/transcribe/segments.go

@@ -0,0 +1,49 @@
+package transcribe
+
+import (
+	"fmt"
+	"strings"
+)
+
+// Segment is one timestamped chunk of a transcript.
+type Segment struct {
+	Start float64 // seconds from start of audio
+	End   float64
+	Text  string
+}
+
+// PlainText joins all segments into a single transcript.
+func PlainText(segs []Segment) string {
+	var b strings.Builder
+	for _, s := range segs {
+		b.WriteString(strings.TrimSpace(s.Text))
+		b.WriteByte(' ')
+	}
+	return strings.TrimSpace(b.String())
+}
+
+// FormatForLLM renders segments as one timestamped line each, suitable for
+// feeding to a model that needs to pick a time window.
+//
+//	[mm:ss] [mm:ss] text
+func FormatForLLM(segs []Segment) string {
+	var b strings.Builder
+	for _, s := range segs {
+		fmt.Fprintf(&b, "[%s] [%s] %s\n", formatTS(s.Start), formatTS(s.End), strings.TrimSpace(s.Text))
+	}
+	return b.String()
+}
+
+func formatTS(seconds float64) string {
+	if seconds < 0 {
+		seconds = 0
+	}
+	total := int(seconds)
+	h := total / 3600
+	m := (total % 3600) / 60
+	s := total % 60
+	if h > 0 {
+		return fmt.Sprintf("%02d:%02d:%02d", h, m, s)
+	}
+	return fmt.Sprintf("%02d:%02d", m, s)
+}