Commitok összehasonlítása
12 Commit-ok
8e1c7bd813
...
main
| Szerző | SHA1 | Dátum | |
|---|---|---|---|
|
|
29d0c0e503 | ||
|
|
fdf8bf9743 | ||
|
|
aa241f55c8 | ||
|
|
913ce5ca1a | ||
|
|
636a4acf97 | ||
|
|
4c680f9e4d | ||
|
|
21be80ddb4 | ||
|
|
865b194eeb | ||
|
|
bd20988fa0 | ||
|
|
3e96a6e394 | ||
|
|
9cd0d3eaa5 | ||
|
|
b0533f85f0 |
1
.gitignore
vendored
1
.gitignore
vendored
@@ -3,6 +3,7 @@ macos/build/
|
||||
macos/WhisperDictate.app/Contents/MacOS/
|
||||
macos/WhisperDictate.app/Contents/Frameworks/
|
||||
macos/WhisperDictate.app/Contents/_CodeSignature/
|
||||
macos/WhisperDictate.app/Contents/CodeResources
|
||||
WhisperDictate
|
||||
*.o
|
||||
*.dSYM/
|
||||
|
||||
17
README.md
17
README.md
@@ -12,17 +12,26 @@ A simple menu bar app for voice dictation using OpenAI Whisper (local, offline).
|
||||
|
||||
## macOS
|
||||
|
||||
### What's new in 1.3.0
|
||||
|
||||
- ⚡ **Warm transcription server** — the Whisper model stays loaded in a background `whisper-server`, so each dictation is a fast local call (~0.5s) instead of reloading the 1.5–1.6 GB model on every utterance
|
||||
- 📊 **Live mic-level HUD** while recording, which warns ("No audio input?") if the microphone stays silent
|
||||
- 📖 **Custom dictionary** — an editable *Heard → Replacement* table that fixes names/jargon in the transcript and also biases recognition
|
||||
|
||||
### Features
|
||||
|
||||
- 🎤 Global hotkey (⌃⌥D) to start/stop recording
|
||||
- ⚡ Fast transcription via a warm whisper-server (model stays loaded — no per-dictation reload), with automatic fallback to `whisper-cli`
|
||||
- 📊 Live mic-level meter while recording, with a "no audio input" warning
|
||||
- 📖 Custom dictionary (heard → replacement) that also biases recognition
|
||||
- 🔒 Fully offline - uses local Whisper model
|
||||
- ⚡ Automatic paste into any focused app
|
||||
- ⌨️ Automatic paste into any focused app
|
||||
- 📋 Clipboard preservation - your copied content is restored after paste
|
||||
- ⚙️ Settings window with model selection dropdown
|
||||
- 📥 Built-in model downloader with progress indicator
|
||||
- 🚀 Launch at login support
|
||||
- 🔊 Sound feedback (optional)
|
||||
- 📦 Self-contained - whisper-cli bundled in app
|
||||
- 📦 Self-contained - whisper-cli and whisper-server bundled in the app
|
||||
|
||||
### Requirements
|
||||
|
||||
@@ -75,6 +84,7 @@ make install
|
||||
Click the menu bar icon → Settings to configure:
|
||||
- **Language**: Auto-detect or 31 supported languages
|
||||
- **Model**: Select from installed models or download new ones
|
||||
- **Dictionary**: Add *Heard → Replacement* pairs to fix names/jargon (use `\n` for a newline); the replacement terms also improve recognition
|
||||
- **Sound feedback**: Toggle audio feedback on/off
|
||||
- **Launch at login**: Start automatically when you log in
|
||||
|
||||
@@ -93,6 +103,8 @@ Download models directly from the app or manually:
|
||||
|
||||
Models are stored in `~/.whisper-models/`
|
||||
|
||||
> With the warm server (1.3.0+), the selected model loads once at startup (a few seconds); after that each dictation skips the reload, so the speeds above are roughly the per-dictation transcription time.
|
||||
|
||||
### Audio Feedback
|
||||
|
||||
- 🔔 **Tink** - Recording started
|
||||
@@ -114,6 +126,7 @@ Grant these in System Settings → Privacy & Security:
|
||||
## Security
|
||||
|
||||
- All processing is done locally - no data leaves your device
|
||||
- The bundled `whisper-server` is bound to `127.0.0.1` (loopback only) and is never exposed on the network
|
||||
- Audio files are stored in private temp directory and deleted after transcription
|
||||
- Clipboard is cleared after paste (transcript doesn't remain accessible)
|
||||
- Original clipboard content is preserved and restored after paste
|
||||
|
||||
@@ -2,7 +2,7 @@
|
||||
|
||||
APP_NAME = WhisperDictate
|
||||
APP_BUNDLE = $(APP_NAME).app
|
||||
VERSION = 1.2.0
|
||||
VERSION = 1.3.0
|
||||
|
||||
# Directories
|
||||
SRC_DIR = src
|
||||
|
||||
@@ -4,6 +4,8 @@
|
||||
<dict>
|
||||
<key>CFBundleDevelopmentRegion</key>
|
||||
<string>en</string>
|
||||
<key>CFBundleDisplayName</key>
|
||||
<string>WhisperDictate</string>
|
||||
<key>CFBundleExecutable</key>
|
||||
<string>WhisperDictate</string>
|
||||
<key>CFBundleIconFile</key>
|
||||
@@ -14,23 +16,26 @@
|
||||
<string>6.0</string>
|
||||
<key>CFBundleName</key>
|
||||
<string>WhisperDictate</string>
|
||||
<key>CFBundleDisplayName</key>
|
||||
<string>WhisperDictate</string>
|
||||
<key>CFBundlePackageType</key>
|
||||
<string>APPL</string>
|
||||
<key>CFBundleShortVersionString</key>
|
||||
<string>1.0.0</string>
|
||||
<string>1.3.0</string>
|
||||
<key>CFBundleVersion</key>
|
||||
<string>1</string>
|
||||
<string>3</string>
|
||||
<key>LSMinimumSystemVersion</key>
|
||||
<string>13.0</string>
|
||||
<key>LSUIElement</key>
|
||||
<true/>
|
||||
<key>NSAppTransportSecurity</key>
|
||||
<dict>
|
||||
<key>NSAllowsLocalNetworking</key>
|
||||
<true/>
|
||||
</dict>
|
||||
<key>NSAppleEventsUsageDescription</key>
|
||||
<string>WhisperDictate needs accessibility access to paste transcribed text.</string>
|
||||
<key>NSHighResolutionCapable</key>
|
||||
<true/>
|
||||
<key>NSMicrophoneUsageDescription</key>
|
||||
<string>WhisperDictate needs microphone access to record your voice for transcription.</string>
|
||||
<key>NSAppleEventsUsageDescription</key>
|
||||
<string>WhisperDictate needs accessibility access to paste transcribed text.</string>
|
||||
</dict>
|
||||
</plist>
|
||||
|
||||
28
macos/WhisperDictate.entitlements
Normal file
28
macos/WhisperDictate.entitlements
Normal file
@@ -0,0 +1,28 @@
|
||||
<?xml version="1.0" encoding="UTF-8"?>
|
||||
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
|
||||
<plist version="1.0">
|
||||
<dict>
|
||||
<!-- Not sandboxed: WhisperDictate uses global Accessibility / synthetic
|
||||
keystrokes and spawns a bundled CLI, which the App Sandbox forbids.
|
||||
Distributed via Developer ID, not the Mac App Store. -->
|
||||
<key>com.apple.security.app-sandbox</key>
|
||||
<false/>
|
||||
<!-- The app bundles whisper-cli + ggml/whisper dylibs in the app. Disabling
|
||||
library validation lets the process tree load them without tripping the
|
||||
hardened runtime's same-Team-ID check. -->
|
||||
<key>com.apple.security.cs.disable-library-validation</key>
|
||||
<true/>
|
||||
<!-- Hardened-runtime resource-access entitlement. REQUIRED for microphone
|
||||
access under the hardened runtime: without it tccd refuses to even
|
||||
prompt and denies kTCCServiceMicrophone outright, so the app records
|
||||
silence (and whisper hallucinates on the empty audio). Must be present
|
||||
on any Developer ID / notarized build. -->
|
||||
<key>com.apple.security.device.audio-input</key>
|
||||
<true/>
|
||||
<!-- whisper.cpp's Metal/ggml backend generates code at runtime; the
|
||||
hardened runtime needs allow-jit so the bundled whisper-cli/dylibs
|
||||
don't get killed for JIT. -->
|
||||
<key>com.apple.security.cs.allow-jit</key>
|
||||
<true/>
|
||||
</dict>
|
||||
</plist>
|
||||
@@ -25,6 +25,17 @@ mkdir -p "$FRAMEWORKS_DIR"
|
||||
cp "$WHISPER_CLI_REAL" "$MACOS_DIR/whisper-cli"
|
||||
chmod +x "$MACOS_DIR/whisper-cli"
|
||||
|
||||
# Copy whisper-server (warm, preloaded-model HTTP backend) from the same bin dir
|
||||
WHISPER_BIN_DIR=$(dirname "$WHISPER_CLI_REAL")
|
||||
WHISPER_SERVER_REAL="$WHISPER_BIN_DIR/whisper-server"
|
||||
if [ -f "$WHISPER_SERVER_REAL" ]; then
|
||||
cp "$WHISPER_SERVER_REAL" "$MACOS_DIR/whisper-server"
|
||||
chmod +x "$MACOS_DIR/whisper-server"
|
||||
echo "Copied: whisper-server"
|
||||
else
|
||||
echo "Warning: whisper-server not found at $WHISPER_SERVER_REAL"
|
||||
fi
|
||||
|
||||
# List of dylibs to copy
|
||||
DYLIBS=(
|
||||
"libwhisper.1.dylib"
|
||||
@@ -49,9 +60,12 @@ for dylib in "${DYLIBS[@]}"; do
|
||||
fi
|
||||
done
|
||||
|
||||
# Fix dylib paths in whisper-cli
|
||||
for dylib in "${DYLIBS[@]}"; do
|
||||
install_name_tool -change "@rpath/$dylib" "@executable_path/../Frameworks/$dylib" "$MACOS_DIR/whisper-cli" 2>/dev/null || true
|
||||
# Fix dylib paths in the bundled executables (whisper-cli + whisper-server)
|
||||
for exe in whisper-cli whisper-server; do
|
||||
[ -f "$MACOS_DIR/$exe" ] || continue
|
||||
for dylib in "${DYLIBS[@]}"; do
|
||||
install_name_tool -change "@rpath/$dylib" "@executable_path/../Frameworks/$dylib" "$MACOS_DIR/$exe" 2>/dev/null || true
|
||||
done
|
||||
done
|
||||
|
||||
# Fix dylib paths in each dylib (they reference each other)
|
||||
@@ -67,8 +81,10 @@ for dylib in "${DYLIBS[@]}"; do
|
||||
fi
|
||||
done
|
||||
|
||||
# Sign everything
|
||||
codesign --force --sign - "$MACOS_DIR/whisper-cli" 2>/dev/null || true
|
||||
# Sign everything (ad-hoc; the release flow re-signs with Developer ID afterward)
|
||||
for exe in whisper-cli whisper-server; do
|
||||
[ -f "$MACOS_DIR/$exe" ] && codesign --force --sign - "$MACOS_DIR/$exe" 2>/dev/null || true
|
||||
done
|
||||
for dylib in "${DYLIBS[@]}"; do
|
||||
codesign --force --sign - "$FRAMEWORKS_DIR/$dylib" 2>/dev/null || true
|
||||
done
|
||||
|
||||
@@ -3,12 +3,14 @@ import AVFoundation
|
||||
import Carbon.HIToolbox
|
||||
import ServiceManagement
|
||||
import ApplicationServices
|
||||
import Darwin
|
||||
|
||||
// MARK: - User Defaults Keys
|
||||
struct Defaults {
|
||||
static let language = "whisperLanguage"
|
||||
static let modelPath = "whisperModelPath"
|
||||
static let playSounds = "playSounds"
|
||||
static let replacements = "customReplacements"
|
||||
}
|
||||
|
||||
// MARK: - Whisper Models
|
||||
@@ -150,12 +152,250 @@ struct SupportedLanguages {
|
||||
}
|
||||
}
|
||||
|
||||
// MARK: - Whisper Server (warm, preloaded-model backend)
|
||||
// Keeps a whisper-server child process alive with the model already loaded, so
|
||||
// each dictation is a fast local HTTP call (~0.5s) instead of a multi-second
|
||||
// model + Metal reload on every utterance (what plain whisper-cli does).
|
||||
final class WhisperServer {
|
||||
private let serverPath: String
|
||||
private let modelPath: String
|
||||
private let threads: Int
|
||||
private var process: Process?
|
||||
private var stopping = false
|
||||
private(set) var port: Int = 0
|
||||
private(set) var isReady = false
|
||||
|
||||
init(serverPath: String, modelPath: String, threads: Int) {
|
||||
self.serverPath = serverPath
|
||||
self.modelPath = modelPath
|
||||
self.threads = threads
|
||||
}
|
||||
|
||||
func start(onReady: @escaping (Bool) -> Void) {
|
||||
port = WhisperServer.findFreePort()
|
||||
let task = Process()
|
||||
task.executableURL = URL(fileURLWithPath: serverPath)
|
||||
task.arguments = [
|
||||
"-m", modelPath,
|
||||
"--host", "127.0.0.1",
|
||||
"--port", String(port),
|
||||
"-t", String(threads),
|
||||
"-nt", // no timestamps -> plain text response
|
||||
"-sns", // suppress non-speech tokens -> fewer hallucinations
|
||||
]
|
||||
task.standardOutput = FileHandle.nullDevice
|
||||
task.standardError = FileHandle.nullDevice
|
||||
task.terminationHandler = { [weak self] _ in self?.isReady = false }
|
||||
do {
|
||||
try task.run()
|
||||
process = task
|
||||
} catch {
|
||||
NSLog("WhisperServer: failed to launch: \(error)")
|
||||
onReady(false)
|
||||
return
|
||||
}
|
||||
// Model load can take ~10s (incl. Metal init); poll the port off-thread.
|
||||
DispatchQueue.global(qos: .userInitiated).async { [weak self] in
|
||||
guard let self = self else { return }
|
||||
let deadline = Date().addingTimeInterval(120)
|
||||
while Date() < deadline {
|
||||
if self.stopping || !(self.process?.isRunning ?? false) { break }
|
||||
if WhisperServer.canConnect(port: self.port) {
|
||||
self.isReady = true
|
||||
onReady(true)
|
||||
return
|
||||
}
|
||||
Thread.sleep(forTimeInterval: 0.3)
|
||||
}
|
||||
onReady(false)
|
||||
}
|
||||
}
|
||||
|
||||
func stop() {
|
||||
stopping = true
|
||||
isReady = false
|
||||
process?.terminationHandler = nil
|
||||
process?.terminate()
|
||||
process = nil
|
||||
}
|
||||
|
||||
// Synchronous transcription via the warm server. Call OFF the main thread.
|
||||
func transcribe(wavPath: String, language: String, prompt: String?) -> String? {
|
||||
guard isReady,
|
||||
let url = URL(string: "http://127.0.0.1:\(port)/inference"),
|
||||
let wavData = try? Data(contentsOf: URL(fileURLWithPath: wavPath)) else { return nil }
|
||||
|
||||
let boundary = "----WhisperDictate\(UInt32.random(in: 0 ... .max))"
|
||||
var req = URLRequest(url: url)
|
||||
req.httpMethod = "POST"
|
||||
req.setValue("multipart/form-data; boundary=\(boundary)", forHTTPHeaderField: "Content-Type")
|
||||
req.timeoutInterval = 120
|
||||
|
||||
var body = Data()
|
||||
let nl = "\r\n"
|
||||
func raw(_ s: String) { body.append(s.data(using: .utf8)!) }
|
||||
// audio file part
|
||||
raw("--\(boundary)\(nl)")
|
||||
raw("Content-Disposition: form-data; name=\"file\"; filename=\"audio.wav\"\(nl)")
|
||||
raw("Content-Type: audio/wav\(nl)\(nl)")
|
||||
body.append(wavData)
|
||||
raw(nl)
|
||||
// text fields
|
||||
func field(_ name: String, _ value: String) {
|
||||
raw("--\(boundary)\(nl)")
|
||||
raw("Content-Disposition: form-data; name=\"\(name)\"\(nl)\(nl)")
|
||||
raw("\(value)\(nl)")
|
||||
}
|
||||
field("response_format", "json")
|
||||
field("temperature", "0")
|
||||
if language.lowercased() != SupportedLanguages.autoDetect { field("language", language.lowercased()) }
|
||||
if let prompt = prompt, !prompt.isEmpty { field("prompt", prompt) }
|
||||
raw("--\(boundary)--\(nl)")
|
||||
req.httpBody = body
|
||||
|
||||
let sem = DispatchSemaphore(value: 0)
|
||||
var result: String?
|
||||
URLSession.shared.dataTask(with: req) { data, _, error in
|
||||
defer { sem.signal() }
|
||||
if let error = error { NSLog("WhisperServer: request error: \(error)"); return }
|
||||
guard let data = data,
|
||||
let obj = try? JSONSerialization.jsonObject(with: data) as? [String: Any],
|
||||
let text = obj["text"] as? String else { return }
|
||||
result = text.trimmingCharacters(in: .whitespacesAndNewlines)
|
||||
}.resume()
|
||||
sem.wait()
|
||||
return result
|
||||
}
|
||||
|
||||
// MARK: socket helpers (find a free loopback port + readiness probe)
|
||||
static func findFreePort() -> Int {
|
||||
let fd = socket(AF_INET, SOCK_STREAM, 0)
|
||||
if fd < 0 { return 8123 }
|
||||
defer { close(fd) }
|
||||
var addr = sockaddr_in()
|
||||
addr.sin_family = sa_family_t(AF_INET)
|
||||
addr.sin_addr.s_addr = inet_addr("127.0.0.1")
|
||||
addr.sin_port = 0
|
||||
let bound = withUnsafePointer(to: &addr) {
|
||||
$0.withMemoryRebound(to: sockaddr.self, capacity: 1) {
|
||||
Darwin.bind(fd, $0, socklen_t(MemoryLayout<sockaddr_in>.size))
|
||||
}
|
||||
}
|
||||
guard bound == 0 else { return 8123 }
|
||||
var len = socklen_t(MemoryLayout<sockaddr_in>.size)
|
||||
_ = withUnsafeMutablePointer(to: &addr) {
|
||||
$0.withMemoryRebound(to: sockaddr.self, capacity: 1) {
|
||||
getsockname(fd, $0, &len)
|
||||
}
|
||||
}
|
||||
return Int(UInt16(bigEndian: addr.sin_port))
|
||||
}
|
||||
|
||||
static func canConnect(port: Int) -> Bool {
|
||||
let fd = socket(AF_INET, SOCK_STREAM, 0)
|
||||
if fd < 0 { return false }
|
||||
defer { close(fd) }
|
||||
var addr = sockaddr_in()
|
||||
addr.sin_family = sa_family_t(AF_INET)
|
||||
addr.sin_port = in_port_t(UInt16(port)).bigEndian
|
||||
addr.sin_addr.s_addr = inet_addr("127.0.0.1")
|
||||
let r = withUnsafePointer(to: &addr) {
|
||||
$0.withMemoryRebound(to: sockaddr.self, capacity: 1) {
|
||||
Darwin.connect(fd, $0, socklen_t(MemoryLayout<sockaddr_in>.size))
|
||||
}
|
||||
}
|
||||
return r == 0
|
||||
}
|
||||
}
|
||||
|
||||
// MARK: - Recording HUD (live mic level + silence warning)
|
||||
// A small floating, non-interactive panel shown while recording. Doubles as a
|
||||
// diagnostic: if the input stays silent it flags "no input" — exactly the
|
||||
// failure that silently produced empty recordings + whisper hallucinations.
|
||||
final class MeterView: NSView {
|
||||
private var level: CGFloat = 0
|
||||
private var silent = false
|
||||
|
||||
func set(level: CGFloat, silent: Bool) {
|
||||
self.level = max(0, min(1, level))
|
||||
self.silent = silent
|
||||
needsDisplay = true
|
||||
}
|
||||
func reset() { level = 0; silent = false; needsDisplay = true }
|
||||
|
||||
override func draw(_ dirtyRect: NSRect) {
|
||||
let bg = NSBezierPath(roundedRect: bounds, xRadius: 12, yRadius: 12)
|
||||
NSColor(calibratedWhite: 0.10, alpha: 0.93).setFill()
|
||||
bg.fill()
|
||||
|
||||
let label = silent ? "🎤 No audio input?" : "🎤 Recording…"
|
||||
label.draw(at: NSPoint(x: 16, y: bounds.height - 26), withAttributes: [
|
||||
.font: NSFont.systemFont(ofSize: 12, weight: .medium),
|
||||
.foregroundColor: silent ? NSColor.systemRed : NSColor.white
|
||||
])
|
||||
|
||||
let segs = 20
|
||||
let margin: CGFloat = 16, gap: CGFloat = 3, segH: CGFloat = 14, y: CGFloat = 14
|
||||
let segW = (bounds.width - margin * 2 - gap * CGFloat(segs - 1)) / CGFloat(segs)
|
||||
let lit = Int((level * CGFloat(segs)).rounded())
|
||||
for i in 0..<segs {
|
||||
let rect = NSRect(x: margin + CGFloat(i) * (segW + gap), y: y, width: segW, height: segH)
|
||||
let path = NSBezierPath(roundedRect: rect, xRadius: 2, yRadius: 2)
|
||||
if i < lit {
|
||||
let frac = CGFloat(i) / CGFloat(segs)
|
||||
(silent ? NSColor.systemRed
|
||||
: frac > 0.85 ? .systemRed
|
||||
: frac > 0.6 ? .systemYellow : .systemGreen).setFill()
|
||||
} else {
|
||||
NSColor(calibratedWhite: 1, alpha: 0.12).setFill()
|
||||
}
|
||||
path.fill()
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
final class RecordingHUD {
|
||||
private var window: NSWindow?
|
||||
private let meter = MeterView(frame: NSRect(x: 0, y: 0, width: 260, height: 72))
|
||||
|
||||
func show() {
|
||||
if window == nil {
|
||||
let w = NSPanel(contentRect: NSRect(x: 0, y: 0, width: 260, height: 72),
|
||||
styleMask: [.borderless, .nonactivatingPanel],
|
||||
backing: .buffered, defer: false)
|
||||
w.isOpaque = false
|
||||
w.backgroundColor = .clear
|
||||
w.hasShadow = true
|
||||
w.level = .statusBar
|
||||
w.ignoresMouseEvents = true
|
||||
w.isFloatingPanel = true
|
||||
w.collectionBehavior = [.canJoinAllSpaces, .stationary, .fullScreenAuxiliary, .ignoresCycle]
|
||||
w.contentView = meter
|
||||
window = w
|
||||
}
|
||||
if let screen = NSScreen.main, let window = window {
|
||||
let vf = screen.visibleFrame
|
||||
window.setFrameOrigin(NSPoint(x: vf.midX - window.frame.width / 2, y: vf.minY + 120))
|
||||
}
|
||||
meter.reset()
|
||||
window?.orderFrontRegardless()
|
||||
}
|
||||
|
||||
func update(level: CGFloat, silent: Bool) { meter.set(level: level, silent: silent) }
|
||||
func hide() { window?.orderOut(nil) }
|
||||
}
|
||||
|
||||
// MARK: - App Delegate
|
||||
class AppDelegate: NSObject, NSApplicationDelegate, URLSessionDownloadDelegate {
|
||||
class AppDelegate: NSObject, NSApplicationDelegate, URLSessionDownloadDelegate, NSTableViewDataSource, NSTableViewDelegate, NSTextFieldDelegate, NSWindowDelegate {
|
||||
var statusItem: NSStatusItem!
|
||||
var audioRecorder: AVAudioRecorder?
|
||||
var isRecording = false
|
||||
var settingsWindowController: NSWindowController?
|
||||
var whisperServer: WhisperServer?
|
||||
let recordingHUD = RecordingHUD()
|
||||
var meterTimer: Timer?
|
||||
var silenceStart: Date?
|
||||
weak var dictTable: NSTableView?
|
||||
|
||||
// Use private temp directory with unique filename
|
||||
var audioFilePath: String {
|
||||
@@ -178,19 +418,51 @@ class AppDelegate: NSObject, NSApplicationDelegate, URLSessionDownloadDelegate {
|
||||
set { UserDefaults.standard.set(newValue, forKey: Defaults.playSounds) }
|
||||
}
|
||||
|
||||
// Custom dictionary: ordered list of ["from": recognised, "to": replacement].
|
||||
var replacements: [[String: String]] {
|
||||
get { (UserDefaults.standard.array(forKey: Defaults.replacements) as? [[String: String]]) ?? [] }
|
||||
set { UserDefaults.standard.set(newValue, forKey: Defaults.replacements) }
|
||||
}
|
||||
|
||||
// Whisper initial prompt built from the desired ("to") terms, to bias
|
||||
// recognition toward the user's names/jargon. Returns nil if empty.
|
||||
func customPrompt() -> String? {
|
||||
let terms = replacements.compactMap { $0["to"] }
|
||||
.map { $0.replacingOccurrences(of: "\\n", with: " ") }
|
||||
.map { $0.trimmingCharacters(in: .whitespaces) }
|
||||
.filter { !$0.isEmpty }
|
||||
return terms.isEmpty ? nil : terms.joined(separator: ", ")
|
||||
}
|
||||
|
||||
// Apply the dictionary find/replace pairs to a transcript. "\n" in a
|
||||
// replacement becomes a real newline (e.g. spoken-command → line break).
|
||||
func applyReplacements(_ text: String) -> String {
|
||||
var result = text
|
||||
for pair in replacements {
|
||||
guard let from = pair["from"], !from.isEmpty else { continue }
|
||||
let to = (pair["to"] ?? "").replacingOccurrences(of: "\\n", with: "\n")
|
||||
result = result.replacingOccurrences(of: from, with: to, options: [.caseInsensitive])
|
||||
}
|
||||
return result
|
||||
}
|
||||
|
||||
func applicationDidFinishLaunching(_ notification: Notification) {
|
||||
setupStatusItem()
|
||||
registerHotkey()
|
||||
requestMicrophonePermission()
|
||||
checkAccessibilityPermission()
|
||||
|
||||
// First-run: check if model exists, if not show setup wizard
|
||||
// Start warming the model server BEFORE the (blocking) accessibility
|
||||
// prompt, so the multi-second model load overlaps with the user
|
||||
// granting permissions instead of running after it.
|
||||
if !hasAnyModel() {
|
||||
showFirstRunWizard()
|
||||
} else {
|
||||
checkModelExists()
|
||||
startWhisperServer()
|
||||
}
|
||||
|
||||
checkAccessibilityPermission()
|
||||
|
||||
NSLog("WhisperDictate started. Press ⌃⌥D to toggle recording.")
|
||||
}
|
||||
|
||||
@@ -287,6 +559,7 @@ class AppDelegate: NSObject, NSApplicationDelegate, URLSessionDownloadDelegate {
|
||||
updateStatus("Ready - \(model.name)")
|
||||
if playSounds { NSSound(named: "Glass")?.play() }
|
||||
NSLog("Model downloaded: \(model.name)")
|
||||
startWhisperServer()
|
||||
} catch {
|
||||
statusItem.button?.title = "🎤"
|
||||
updateStatus("⚠️ Save failed")
|
||||
@@ -310,6 +583,7 @@ class AppDelegate: NSObject, NSApplicationDelegate, URLSessionDownloadDelegate {
|
||||
}
|
||||
|
||||
func applicationWillTerminate(_ notification: Notification) {
|
||||
whisperServer?.stop()
|
||||
cleanupAudioFile()
|
||||
}
|
||||
|
||||
@@ -350,17 +624,18 @@ class AppDelegate: NSObject, NSApplicationDelegate, URLSessionDownloadDelegate {
|
||||
|
||||
func createSettingsWindow() -> NSWindow {
|
||||
let window = NSWindow(
|
||||
contentRect: NSRect(x: 0, y: 0, width: 450, height: 280),
|
||||
contentRect: NSRect(x: 0, y: 0, width: 450, height: 540),
|
||||
styleMask: [.titled, .closable],
|
||||
backing: .buffered,
|
||||
defer: false
|
||||
)
|
||||
window.title = "WhisperDictate Settings"
|
||||
window.center()
|
||||
window.delegate = self
|
||||
|
||||
let contentView = NSView(frame: window.contentView!.bounds)
|
||||
|
||||
var y: CGFloat = 230
|
||||
var y: CGFloat = 498
|
||||
let labelWidth: CGFloat = 120
|
||||
let controlX: CGFloat = 140
|
||||
let controlWidth: CGFloat = 280
|
||||
@@ -453,6 +728,44 @@ class AppDelegate: NSObject, NSApplicationDelegate, URLSessionDownloadDelegate {
|
||||
loginCheck.state = isLaunchAtLoginEnabled() ? .on : .off
|
||||
contentView.addSubview(loginCheck)
|
||||
|
||||
// Custom dictionary (recognised -> replacement)
|
||||
let dictLabel = NSTextField(labelWithString: "Dictionary (heard → replacement):")
|
||||
dictLabel.frame = NSRect(x: 20, y: 304, width: 300, height: 20)
|
||||
contentView.addSubview(dictLabel)
|
||||
|
||||
let scroll = NSScrollView(frame: NSRect(x: 20, y: 104, width: 410, height: 192))
|
||||
scroll.hasVerticalScroller = true
|
||||
scroll.borderType = .bezelBorder
|
||||
let table = NSTableView(frame: scroll.bounds)
|
||||
table.usesAlternatingRowBackgroundColors = true
|
||||
table.rowHeight = 22
|
||||
let colFrom = NSTableColumn(identifier: NSUserInterfaceItemIdentifier("from"))
|
||||
colFrom.title = "Heard"; colFrom.width = 195; colFrom.isEditable = true
|
||||
let colTo = NSTableColumn(identifier: NSUserInterfaceItemIdentifier("to"))
|
||||
colTo.title = "Replacement"; colTo.width = 195
|
||||
table.addTableColumn(colFrom)
|
||||
table.addTableColumn(colTo)
|
||||
table.dataSource = self
|
||||
table.delegate = self
|
||||
scroll.documentView = table
|
||||
contentView.addSubview(scroll)
|
||||
dictTable = table
|
||||
|
||||
let addBtn = NSButton(title: "+", target: self, action: #selector(addReplacement))
|
||||
addBtn.frame = NSRect(x: 20, y: 70, width: 34, height: 26)
|
||||
addBtn.bezelStyle = .rounded
|
||||
contentView.addSubview(addBtn)
|
||||
let delBtn = NSButton(title: "–", target: self, action: #selector(removeReplacement))
|
||||
delBtn.frame = NSRect(x: 56, y: 70, width: 34, height: 26)
|
||||
delBtn.bezelStyle = .rounded
|
||||
contentView.addSubview(delBtn)
|
||||
|
||||
let dictHint = NSTextField(labelWithString: "The Replacement column also improves recognition. \\n = newline.")
|
||||
dictHint.frame = NSRect(x: 100, y: 74, width: 330, height: 18)
|
||||
dictHint.font = NSFont.systemFont(ofSize: 10)
|
||||
dictHint.textColor = .tertiaryLabelColor
|
||||
contentView.addSubview(dictHint)
|
||||
|
||||
// Models directory hint
|
||||
let hintLabel = NSTextField(labelWithString: "Models stored in: ~/.whisper-models/")
|
||||
hintLabel.frame = NSRect(x: 20, y: 15, width: 410, height: 20)
|
||||
@@ -487,6 +800,7 @@ class AppDelegate: NSObject, NSApplicationDelegate, URLSessionDownloadDelegate {
|
||||
let path = installedModels[index].path
|
||||
modelPath = path
|
||||
checkModelExists()
|
||||
startWhisperServer()
|
||||
NSLog("Model changed to: \(path)")
|
||||
}
|
||||
}
|
||||
@@ -549,6 +863,72 @@ class AppDelegate: NSObject, NSApplicationDelegate, URLSessionDownloadDelegate {
|
||||
setLaunchAtLogin(sender.state == .on)
|
||||
}
|
||||
|
||||
// MARK: - Custom dictionary table (view-based, editable text fields)
|
||||
func numberOfRows(in tableView: NSTableView) -> Int { replacements.count }
|
||||
|
||||
func tableView(_ tableView: NSTableView, viewFor tableColumn: NSTableColumn?, row: Int) -> NSView? {
|
||||
guard let column = tableColumn, row < replacements.count else { return nil }
|
||||
let id = column.identifier
|
||||
let field: NSTextField
|
||||
if let reused = tableView.makeView(withIdentifier: id, owner: self) as? NSTextField {
|
||||
field = reused
|
||||
} else {
|
||||
field = NSTextField()
|
||||
field.identifier = id
|
||||
field.isBordered = false
|
||||
field.drawsBackground = false
|
||||
field.isEditable = true
|
||||
field.isSelectable = true
|
||||
field.font = NSFont.systemFont(ofSize: 12)
|
||||
field.lineBreakMode = .byTruncatingTail
|
||||
field.delegate = self
|
||||
field.placeholderString = id.rawValue == "from" ? "heard word" : "replacement (\\n = newline)"
|
||||
}
|
||||
field.stringValue = replacements[row][id.rawValue] ?? ""
|
||||
return field
|
||||
}
|
||||
|
||||
// Commit an edited cell back into the model (fires on Enter, Tab, focus loss).
|
||||
func controlTextDidEndEditing(_ obj: Notification) {
|
||||
guard let field = obj.object as? NSTextField, let table = dictTable else { return }
|
||||
let row = table.row(for: field), col = table.column(for: field)
|
||||
guard row >= 0, row < replacements.count, col >= 0, col < table.tableColumns.count else { return }
|
||||
let id = table.tableColumns[col].identifier.rawValue
|
||||
var rows = replacements
|
||||
rows[row][id] = field.stringValue
|
||||
replacements = rows
|
||||
}
|
||||
|
||||
@objc func addReplacement() {
|
||||
pruneEmptyReplacements()
|
||||
var rows = replacements
|
||||
rows.append(["from": "", "to": ""])
|
||||
replacements = rows
|
||||
dictTable?.reloadData()
|
||||
dictTable?.editColumn(0, row: rows.count - 1, with: nil, select: true)
|
||||
}
|
||||
|
||||
@objc func removeReplacement() {
|
||||
guard let table = dictTable, table.selectedRow >= 0, table.selectedRow < replacements.count else { return }
|
||||
var rows = replacements
|
||||
rows.remove(at: table.selectedRow)
|
||||
replacements = rows
|
||||
table.reloadData()
|
||||
}
|
||||
|
||||
// Drop rows where both columns are blank so empty rows aren't persisted.
|
||||
func pruneEmptyReplacements() {
|
||||
let cleaned = replacements.filter {
|
||||
!(($0["from"] ?? "").trimmingCharacters(in: .whitespaces).isEmpty
|
||||
&& ($0["to"] ?? "").trimmingCharacters(in: .whitespaces).isEmpty)
|
||||
}
|
||||
if cleaned.count != replacements.count { replacements = cleaned }
|
||||
}
|
||||
|
||||
func windowWillClose(_ notification: Notification) {
|
||||
pruneEmptyReplacements()
|
||||
}
|
||||
|
||||
// MARK: - Launch at Login
|
||||
func isLaunchAtLoginEnabled() -> Bool {
|
||||
if #available(macOS 13.0, *) {
|
||||
@@ -698,21 +1078,58 @@ class AppDelegate: NSObject, NSApplicationDelegate, URLSessionDownloadDelegate {
|
||||
|
||||
do {
|
||||
audioRecorder = try AVAudioRecorder(url: audioURL, settings: settings)
|
||||
audioRecorder?.isMeteringEnabled = true
|
||||
audioRecorder?.record()
|
||||
isRecording = true
|
||||
statusItem.button?.title = "🔴"
|
||||
updateStatus("Recording...")
|
||||
if playSounds { NSSound(named: "Tink")?.play() }
|
||||
NSLog("Recording started")
|
||||
startMeter()
|
||||
} catch {
|
||||
NSLog("Recording failed: \(error)")
|
||||
if playSounds { NSSound(named: "Basso")?.play() }
|
||||
}
|
||||
}
|
||||
|
||||
// MARK: - Live mic meter (HUD)
|
||||
func startMeter() {
|
||||
silenceStart = nil
|
||||
recordingHUD.show()
|
||||
meterTimer?.invalidate()
|
||||
meterTimer = Timer.scheduledTimer(withTimeInterval: 0.05, repeats: true) { [weak self] _ in
|
||||
self?.updateMeter()
|
||||
}
|
||||
}
|
||||
|
||||
func stopMeter() {
|
||||
meterTimer?.invalidate()
|
||||
meterTimer = nil
|
||||
silenceStart = nil
|
||||
recordingHUD.hide()
|
||||
}
|
||||
|
||||
func updateMeter() {
|
||||
guard isRecording, let recorder = audioRecorder else { return }
|
||||
recorder.updateMeters()
|
||||
let avg = recorder.averagePower(forChannel: 0) // dBFS, ~-160...0
|
||||
let peak = recorder.peakPower(forChannel: 0)
|
||||
let level = max(0, min(1, CGFloat((avg + 50) / 50)))
|
||||
|
||||
// Flag "no input" if the peak stays at silence for over ~1.2s.
|
||||
if peak < -50 {
|
||||
if silenceStart == nil { silenceStart = Date() }
|
||||
} else {
|
||||
silenceStart = nil
|
||||
}
|
||||
let silent = silenceStart.map { Date().timeIntervalSince($0) > 1.2 } ?? false
|
||||
recordingHUD.update(level: level, silent: silent)
|
||||
}
|
||||
|
||||
func stopRecordingAndTranscribe() {
|
||||
audioRecorder?.stop()
|
||||
isRecording = false
|
||||
stopMeter()
|
||||
statusItem.button?.title = "⏳"
|
||||
updateStatus("Transcribing...")
|
||||
if playSounds { NSSound(named: "Pop")?.play() }
|
||||
@@ -749,9 +1166,57 @@ class AppDelegate: NSObject, NSApplicationDelegate, URLSessionDownloadDelegate {
|
||||
return nil
|
||||
}
|
||||
|
||||
func findWhisperServer() -> String? {
|
||||
if let bundlePath = Bundle.main.executablePath {
|
||||
let bundled = (bundlePath as NSString).deletingLastPathComponent + "/whisper-server"
|
||||
if FileManager.default.isExecutableFile(atPath: bundled) { return bundled }
|
||||
}
|
||||
let paths = [
|
||||
"/opt/homebrew/bin/whisper-server",
|
||||
"/usr/local/bin/whisper-server",
|
||||
"/opt/local/bin/whisper-server",
|
||||
NSHomeDirectory() + "/bin/whisper-server"
|
||||
]
|
||||
for path in paths where FileManager.default.isExecutableFile(atPath: path) { return path }
|
||||
return nil
|
||||
}
|
||||
|
||||
// (Re)start the warm whisper-server for the current model. Falls back to
|
||||
// per-call whisper-cli if the server binary is missing or fails to come up.
|
||||
func startWhisperServer() {
|
||||
whisperServer?.stop()
|
||||
whisperServer = nil
|
||||
|
||||
guard let serverPath = findWhisperServer() else {
|
||||
NSLog("whisper-server not found — using per-call whisper-cli")
|
||||
return
|
||||
}
|
||||
guard isValidModelPath(modelPath).valid else { return }
|
||||
|
||||
statusItem.button?.title = "⏳"
|
||||
updateStatus("Loading model…")
|
||||
|
||||
let threads = min(8, max(2, ProcessInfo.processInfo.activeProcessorCount - 2))
|
||||
let server = WhisperServer(serverPath: serverPath, modelPath: modelPath, threads: threads)
|
||||
whisperServer = server
|
||||
server.start { [weak self] ok in
|
||||
DispatchQueue.main.async {
|
||||
guard let self = self, self.whisperServer === server else { return }
|
||||
self.statusItem.button?.title = "🎤"
|
||||
if ok {
|
||||
self.updateStatus("Ready")
|
||||
NSLog("whisper-server ready on port \(server.port)")
|
||||
} else {
|
||||
self.updateStatus("Ready (CLI fallback)")
|
||||
self.whisperServer = nil
|
||||
NSLog("whisper-server failed to start — falling back to whisper-cli")
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// MARK: - Transcription
|
||||
func transcribe() {
|
||||
// Validate inputs before execution
|
||||
let modelValidation = isValidModelPath(modelPath)
|
||||
guard modelValidation.valid else {
|
||||
DispatchQueue.main.async {
|
||||
@@ -771,6 +1236,34 @@ class AppDelegate: NSObject, NSApplicationDelegate, URLSessionDownloadDelegate {
|
||||
return
|
||||
}
|
||||
|
||||
// Prefer the warm server (~0.5s); fall back to a per-call whisper-cli spawn.
|
||||
let prompt = customPrompt()
|
||||
let result: String?
|
||||
if let server = whisperServer, server.isReady {
|
||||
result = server.transcribe(wavPath: audioFilePath, language: language, prompt: prompt)
|
||||
} else {
|
||||
result = transcribeWithCLI(prompt: prompt)
|
||||
}
|
||||
|
||||
self.cleanupAudioFile()
|
||||
|
||||
let finalText = result.map { applyReplacements($0) }
|
||||
|
||||
DispatchQueue.main.async {
|
||||
if let finalText = finalText, !finalText.isEmpty {
|
||||
self.pasteText(finalText)
|
||||
} else {
|
||||
self.statusItem.button?.title = "🎤"
|
||||
self.updateStatus("Ready")
|
||||
if self.playSounds { NSSound(named: "Basso")?.play() }
|
||||
NSLog("No speech recognized")
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// Per-call transcription via whisper-cli (reloads the model every time).
|
||||
// Fallback for when the warm server is unavailable.
|
||||
func transcribeWithCLI(prompt: String? = nil) -> String? {
|
||||
guard let whisperPath = findWhisperCLI() else {
|
||||
DispatchQueue.main.async {
|
||||
self.statusItem.button?.title = "🎤"
|
||||
@@ -778,17 +1271,16 @@ class AppDelegate: NSObject, NSApplicationDelegate, URLSessionDownloadDelegate {
|
||||
if self.playSounds { NSSound(named: "Basso")?.play() }
|
||||
NSLog("whisper-cli not found in PATH")
|
||||
}
|
||||
return
|
||||
return nil
|
||||
}
|
||||
|
||||
let task = Process()
|
||||
task.executableURL = URL(fileURLWithPath: whisperPath)
|
||||
|
||||
// Build arguments - skip language flag for auto-detect
|
||||
var args = ["-m", modelPath]
|
||||
if language.lowercased() != SupportedLanguages.autoDetect {
|
||||
args += ["-l", language.lowercased()]
|
||||
}
|
||||
if let prompt = prompt, !prompt.isEmpty { args += ["--prompt", prompt] }
|
||||
args += ["-f", audioFilePath]
|
||||
task.arguments = args
|
||||
|
||||
@@ -799,45 +1291,18 @@ class AppDelegate: NSObject, NSApplicationDelegate, URLSessionDownloadDelegate {
|
||||
do {
|
||||
try task.run()
|
||||
task.waitUntilExit()
|
||||
|
||||
let data = pipe.fileHandleForReading.readDataToEndOfFile()
|
||||
let output = String(data: data, encoding: .utf8) ?? ""
|
||||
|
||||
let lines = output.components(separatedBy: "\n")
|
||||
var result = ""
|
||||
for line in lines {
|
||||
if line.hasPrefix("[") {
|
||||
for line in output.components(separatedBy: "\n") where line.hasPrefix("[") {
|
||||
if let range = line.range(of: "]") {
|
||||
let text = String(line[range.upperBound...]).trimmingCharacters(in: .whitespaces)
|
||||
result += text + " "
|
||||
}
|
||||
}
|
||||
}
|
||||
result = result.trimmingCharacters(in: .whitespaces)
|
||||
|
||||
// Cleanup audio file after transcription
|
||||
self.cleanupAudioFile()
|
||||
|
||||
DispatchQueue.main.async {
|
||||
if !result.isEmpty {
|
||||
self.pasteText(result)
|
||||
} else {
|
||||
self.statusItem.button?.title = "🎤"
|
||||
self.updateStatus("Ready")
|
||||
if self.playSounds { NSSound(named: "Basso")?.play() }
|
||||
NSLog("No speech recognized")
|
||||
result += String(line[range.upperBound...]).trimmingCharacters(in: .whitespaces) + " "
|
||||
}
|
||||
}
|
||||
return result.trimmingCharacters(in: .whitespaces)
|
||||
} catch {
|
||||
// Cleanup even on error
|
||||
self.cleanupAudioFile()
|
||||
|
||||
DispatchQueue.main.async {
|
||||
self.statusItem.button?.title = "🎤"
|
||||
self.updateStatus("Error")
|
||||
if self.playSounds { NSSound(named: "Basso")?.play() }
|
||||
NSLog("Transcription failed: \(error)")
|
||||
}
|
||||
return nil
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
Reference in New Issue
Block a user