Grant Gilliam
4c086251d9
Native editor to fix speaker-ID errors after transcription (modeled on recap-relay's correction UX): rename a speaker in the legend, merge two speakers, or reassign an individual transcript line. Saving rewrites speakers.json, re-renders transcript.md + recap.html, and updates the voiceprint memory — so a correction compounds: naming an "Unknown" speaker teaches that voice for future calls. - SpeakerEditing (pure, tested): replaceSpeaker (rename = merge-onto-existing), reassign, netNameMap (compose ops), and remap (apply a name map to a recap's structured fields + whole-word free text, so summaries/extras update without re-LLM). - RecapEditModel (@MainActor): loads speakers.json (+ optional recap.json + cluster_fingerprints.json); on save writes the resolved speakers.json, re-renders, and reconciles voiceprints — merge keeps the survivor's print; rename/name-an-Unknown enrolls the cluster's fingerprint under the new name. - TranscriptEditorView (SwiftUI) + EditorWindow (AppKit window for the LSUIElement app); menu gains "Edit speakers". - Pipeline now persists cluster_fingerprints.json (every cluster incl. Unknown) and recap.json (RecapFile) so the editor can learn voices + re-render offline. - RecapModels made Codable; TranscriptAssembler exposes allFingerprints; VoiceprintStore gains enroll() + merge(). 52/52 XCTest (6 new, incl. a full rename→artifacts→voiceprint round-trip on disk).
84 lines
3.6 KiB
Swift
84 lines
3.6 KiB
Swift
import Foundation
|
|
|
|
/// Concatenates per-chunk `label-merge` results into one global `speakers.json`:
|
|
/// segment times offset back to global seconds, speakers unified across chunks by
|
|
/// name, and fingerprints collected for the voiceprint store.
|
|
enum TranscriptAssembler {
|
|
struct ChunkResult {
|
|
let chunkStart: Double // global seconds
|
|
let response: LabelMergeResponse
|
|
}
|
|
|
|
struct Assembled {
|
|
let speakersFile: SpeakersFile
|
|
let fingerprints: [String: [Float]] // confidently-named only, for VoiceprintStore
|
|
let allFingerprints: [String: [Float]] // EVERY cluster incl. Unknown — for editor voice-learning
|
|
}
|
|
|
|
/// Source ranking when the same name appears across chunks with different sources.
|
|
/// `mic_channel` (the local user's own microphone) is the most authoritative.
|
|
private static func rank(_ source: String) -> Int {
|
|
switch source {
|
|
case "mic_channel": return 4
|
|
case "visual": return 3
|
|
case "voiceprint": return 2
|
|
default: return 1 // unmatched
|
|
}
|
|
}
|
|
|
|
private static func isUnknown(_ name: String) -> Bool {
|
|
LabelMergeResponse.isUnknownName(name)
|
|
}
|
|
|
|
static func assemble(sessionId: String, app: String, chunks: [ChunkResult]) -> Assembled {
|
|
var segments: [SpeakersFile.Segment] = []
|
|
var bestSpeaker: [String: SpeakersFile.Speaker] = [:]
|
|
var fingerprints: [String: [Float]] = [:]
|
|
var allFingerprints: [String: [Float]] = [:]
|
|
var models: [String: String] = [:]
|
|
var duration = 0.0
|
|
|
|
for chunk in chunks {
|
|
let offset = chunk.chunkStart
|
|
// Audio length from the chunk window, so silent/all-unknown calls still
|
|
// report a real duration (not just the last segment's end).
|
|
duration = max(duration, offset + chunk.response.duration)
|
|
|
|
for seg in chunk.response.segments {
|
|
let start = seg.startSeconds + offset
|
|
let end = seg.endSeconds + offset
|
|
segments.append(.init(start: start, end: end, speaker: seg.speaker, text: seg.text))
|
|
duration = max(duration, end)
|
|
}
|
|
|
|
for sp in chunk.response.speakers {
|
|
let candidate = SpeakersFile.Speaker(
|
|
name: sp.name, source: sp.source,
|
|
overlapConfidence: sp.overlapConfidence, matchSimilarity: sp.matchSimilarity)
|
|
if let existing = bestSpeaker[sp.name] {
|
|
if rank(sp.source) > rank(existing.source) { bestSpeaker[sp.name] = candidate }
|
|
} else {
|
|
bestSpeaker[sp.name] = candidate
|
|
}
|
|
if let fp = sp.fingerprint, fp.count > 0 {
|
|
allFingerprints[sp.name] = fp // every cluster, for the editor
|
|
if !isUnknown(sp.name) { fingerprints[sp.name] = fp } // named only, for the store
|
|
}
|
|
}
|
|
for (name, fp) in chunk.response.fingerprints where fp.count > 0 {
|
|
allFingerprints[name] = fp
|
|
if !isUnknown(name) { fingerprints[name] = fp }
|
|
}
|
|
}
|
|
|
|
segments.sort { $0.start < $1.start }
|
|
let speakers = bestSpeaker.values.sorted { $0.name < $1.name }
|
|
models = chunks.last?.response.models ?? [:]
|
|
|
|
let file = SpeakersFile(
|
|
sessionId: sessionId, app: app, durationSec: duration,
|
|
speakers: speakers, segments: segments, models: models)
|
|
return Assembled(speakersFile: file, fingerprints: fingerprints, allFingerprints: allFingerprints)
|
|
}
|
|
}
|