feat: share wake gate via SwabbleKit

This commit is contained in:
Peter Steinberger
2025-12-23 01:30:40 +01:00
parent cf48d297dd
commit ef35868bef
2945 changed files with 27887 additions and 122 deletions

View File

@@ -2,6 +2,7 @@ import AVFAudio
import Foundation
import Observation
import Speech
import SwabbleKit
private func makeAudioTapEnqueueCallback(queue: AudioBufferQueue) -> @Sendable (AVAudioPCMBuffer, AVAudioTime) -> Void {
{ buffer, _ in
@@ -289,15 +290,18 @@ final class VoiceWakeManager: NSObject {
private nonisolated func makeRecognitionResultHandler() -> @Sendable (SFSpeechRecognitionResult?, Error?) -> Void {
{ [weak self] result, error in
let transcript = result?.bestTranscription.formattedString
let segments = result.flatMap { result in
transcript.map { WakeWordSpeechSegments.from(transcription: result.bestTranscription, transcript: $0) }
} ?? []
let errorText = error?.localizedDescription
Task { @MainActor in
self?.handleRecognitionCallback(transcript: transcript, errorText: errorText)
self?.handleRecognitionCallback(transcript: transcript, segments: segments, errorText: errorText)
}
}
}
private func handleRecognitionCallback(transcript: String?, errorText: String?) {
private func handleRecognitionCallback(transcript: String?, segments: [WakeWordSegment], errorText: String?) {
if let errorText {
self.statusText = "Recognizer error: \(errorText)"
self.isListening = false
@@ -313,7 +317,7 @@ final class VoiceWakeManager: NSObject {
}
guard let transcript else { return }
guard let cmd = self.extractCommand(from: transcript) else { return }
guard let cmd = self.extractCommand(from: transcript, segments: segments) else { return }
if cmd == self.lastDispatched { return }
self.lastDispatched = cmd
@@ -334,30 +338,18 @@ final class VoiceWakeManager: NSObject {
}
}
private func extractCommand(from transcript: String) -> String? {
Self.extractCommand(from: transcript, triggers: self.activeTriggerWords)
private func extractCommand(from transcript: String, segments: [WakeWordSegment]) -> String? {
Self.extractCommand(from: transcript, segments: segments, triggers: self.activeTriggerWords)
}
nonisolated static func extractCommand(from transcript: String, triggers: [String]) -> String? {
var bestRange: Range<String.Index>?
for trigger in triggers {
let token = trigger.trimmingCharacters(in: .whitespacesAndNewlines)
guard !token.isEmpty else { continue }
guard let range = transcript.range(of: token, options: [.caseInsensitive, .backwards]) else { continue }
if let currentBest = bestRange {
if range.lowerBound > currentBest.lowerBound {
bestRange = range
}
} else {
bestRange = range
}
}
guard let bestRange else { return nil }
let after = transcript[bestRange.upperBound...]
let trimmed = after.trimmingCharacters(in: .whitespacesAndNewlines)
guard !trimmed.isEmpty else { return nil }
return String(trimmed)
nonisolated static func extractCommand(
from transcript: String,
segments: [WakeWordSegment],
triggers: [String],
minPostTriggerGap: TimeInterval = 0.45) -> String?
{
let config = WakeWordGateConfig(triggers: triggers, minPostTriggerGap: minPostTriggerGap)
return WakeWordGate.match(transcript: transcript, segments: segments, config: config)?.command
}
private static func configureAudioSession() throws {

View File

@@ -54,3 +54,4 @@ Sources/Voice/VoiceWakePreferences.swift
../shared/ClawdisKit/Sources/ClawdisKit/ScreenCommands.swift
../shared/ClawdisKit/Sources/ClawdisKit/StoragePaths.swift
../shared/ClawdisKit/Sources/ClawdisKit/SystemCommands.swift
../../Swabble/Sources/SwabbleKit/WakeWordGate.swift

View File

@@ -1,33 +1,90 @@
import Foundation
import Testing
import SwabbleKit
@testable import Clawdis
@Suite struct VoiceWakeManagerExtractCommandTests {
@Test func extractCommandReturnsNilWhenNoTriggerFound() {
#expect(VoiceWakeManager.extractCommand(from: "hello world", triggers: ["clawd"]) == nil)
let transcript = "hello world"
let segments = makeSegments(
transcript: transcript,
words: [("hello", 0.0, 0.1), ("world", 0.2, 0.1)])
#expect(VoiceWakeManager.extractCommand(from: transcript, segments: segments, triggers: ["clawd"]) == nil)
}
@Test func extractCommandTrimsTokensAndResult() {
let cmd = VoiceWakeManager.extractCommand(from: "hey clawd do thing ", triggers: [" clawd "])
let transcript = "hey clawd do thing"
let segments = makeSegments(
transcript: transcript,
words: [
("hey", 0.0, 0.1),
("clawd", 0.2, 0.1),
("do", 0.9, 0.1),
("thing", 1.1, 0.1),
])
let cmd = VoiceWakeManager.extractCommand(
from: transcript,
segments: segments,
triggers: [" clawd "],
minPostTriggerGap: 0.3)
#expect(cmd == "do thing")
}
@Test func extractCommandPicksLatestTriggerOccurrence() {
let transcript = "clawd first\nthen something\nclaude second"
let cmd = VoiceWakeManager.extractCommand(from: transcript, triggers: ["clawd", "claude"])
#expect(cmd == "second")
}
@Test func extractCommandIsCaseInsensitive() {
let cmd = VoiceWakeManager.extractCommand(from: "HELLO CLAWD run it", triggers: ["clawd"])
#expect(cmd == "run it")
@Test func extractCommandReturnsNilWhenGapTooShort() {
let transcript = "hey clawd do thing"
let segments = makeSegments(
transcript: transcript,
words: [
("hey", 0.0, 0.1),
("clawd", 0.2, 0.1),
("do", 0.35, 0.1),
("thing", 0.5, 0.1),
])
let cmd = VoiceWakeManager.extractCommand(
from: transcript,
segments: segments,
triggers: ["clawd"],
minPostTriggerGap: 0.3)
#expect(cmd == nil)
}
@Test func extractCommandReturnsNilWhenNothingAfterTrigger() {
#expect(VoiceWakeManager.extractCommand(from: "hey clawd \n", triggers: ["clawd"]) == nil)
let transcript = "hey clawd"
let segments = makeSegments(
transcript: transcript,
words: [("hey", 0.0, 0.1), ("clawd", 0.2, 0.1)])
#expect(VoiceWakeManager.extractCommand(from: transcript, segments: segments, triggers: ["clawd"]) == nil)
}
@Test func extractCommandIgnoresEmptyTriggers() {
let cmd = VoiceWakeManager.extractCommand(from: "hey clawd do thing", triggers: ["", " ", "clawd"])
let transcript = "hey clawd do thing"
let segments = makeSegments(
transcript: transcript,
words: [
("hey", 0.0, 0.1),
("clawd", 0.2, 0.1),
("do", 0.9, 0.1),
("thing", 1.1, 0.1),
])
let cmd = VoiceWakeManager.extractCommand(
from: transcript,
segments: segments,
triggers: ["", " ", "clawd"],
minPostTriggerGap: 0.3)
#expect(cmd == "do thing")
}
}
private func makeSegments(
transcript: String,
words: [(String, TimeInterval, TimeInterval)])
-> [WakeWordSegment] {
var searchStart = transcript.startIndex
var output: [WakeWordSegment] = []
for (word, start, duration) in words {
let range = transcript.range(of: word, range: searchStart..<transcript.endIndex)
output.append(WakeWordSegment(text: word, start: start, duration: duration, range: range))
if let range { searchStart = range.upperBound }
}
return output
}

View File

@@ -8,6 +8,8 @@ options:
packages:
ClawdisKit:
path: ../shared/ClawdisKit
Swabble:
path: ../../Swabble
schemes:
Clawdis:
@@ -29,6 +31,8 @@ targets:
- package: ClawdisKit
- package: ClawdisKit
product: ClawdisChatUI
- package: Swabble
product: SwabbleKit
- sdk: AppIntents.framework
preBuildScripts:
- name: SwiftFormat (lint)
@@ -86,6 +90,8 @@ targets:
- path: Tests
dependencies:
- target: Clawdis
- package: Swabble
product: SwabbleKit
settings:
base:
PRODUCT_BUNDLE_IDENTIFIER: com.steipete.clawdis.ios.tests