Add code to pace the text to the GS to reduce the amount of re-typing required.

This commit is contained in:
Jeremy Rand 2022-02-22 23:52:00 -05:00
parent 241190600e
commit b2c2cbd587
3 changed files with 171 additions and 77 deletions

View File

@ -52,7 +52,7 @@ struct GSView: View {
speechForwarder.connect(destination: ipAddress) speechForwarder.connect(destination: ipAddress)
} }
} }
.disabled(false) .disabled(speechForwarder.connecting)
.buttonStyle(GSButtonStyle()) .buttonStyle(GSButtonStyle())
Button(speechForwarder.listening ? Button(speechForwarder.listening ?
@ -60,7 +60,7 @@ struct GSView: View {
"\(Image(systemName: "ear.and.waveform")) Listen and Send Text") { "\(Image(systemName: "ear.and.waveform")) Listen and Send Text") {
speechForwarder.listen() speechForwarder.listen()
} }
.disabled(!speechForwarder.connected) .disabled((!speechForwarder.connected) || (!speechForwarder.listening && speechForwarder.sending))
.buttonStyle(GSButtonStyle()) .buttonStyle(GSButtonStyle())
} }
.fixedSize(horizontal: true, vertical: false) .fixedSize(horizontal: true, vertical: false)

View File

@ -19,7 +19,7 @@
<key>CFBundleShortVersionString</key> <key>CFBundleShortVersionString</key>
<string>1.0</string> <string>1.0</string>
<key>CFBundleVersion</key> <key>CFBundleVersion</key>
<string>390</string> <string>434</string>
<key>LSApplicationCategoryType</key> <key>LSApplicationCategoryType</key>
<string>public.app-category.utilities</string> <string>public.app-category.utilities</string>
<key>LSRequiresIPhoneOS</key> <key>LSRequiresIPhoneOS</key>

View File

@ -12,10 +12,13 @@ import Speech
class SpeechForwarder : ObservableObject { class SpeechForwarder : ObservableObject {
@Published var listening = false @Published var listening = false
@Published var connected = false @Published var connected = false
@Published var connecting = false
@Published var textHeard = "" @Published var textHeard = ""
@Published var sending = false
let LISTEN_STATE_MSG = 1 let LISTEN_STATE_MSG = 1
let LISTEN_TEXT_MSG = 2 let LISTEN_TEXT_MSG = 2
let LISTEN_SEND_MORE = 3
let port = 19026 let port = 19026
private var client: TCPClient? private var client: TCPClient?
@ -30,19 +33,31 @@ class SpeechForwarder : ObservableObject {
private let logger = Logger() private let logger = Logger()
private let queue = OperationQueue()
private var condition = NSCondition()
private var latestText = ""
func connect(destination : String) { func connect(destination : String) {
logger.debug("Attempting to connect to \(destination)") connecting = true
client = TCPClient(address: destination, port: Int32(port)) queue.addOperation {
guard let client = client else { return } self.logger.debug("Attempting to connect to \(destination)")
switch client.connect(timeout: 10) { self.client = TCPClient(address: destination, port: Int32(self.port))
case .success: guard let client = self.client else {
connected = true OperationQueue.main.addOperation { self.connecting = false }
logger.debug("Connected to \(destination)") return
case .failure(let error): }
client.close() switch client.connect(timeout: 10) {
self.client = nil case .success:
logger.error("Failed to connect to \(destination): \(String(describing: error))") OperationQueue.main.addOperation { self.connected = true }
break self.logger.debug("Connected to \(destination)")
case .failure(let error):
client.close()
self.client = nil
self.logger.error("Failed to connect to \(destination): \(String(describing: error))")
break
}
OperationQueue.main.addOperation { self.connecting = false }
} }
} }
@ -52,7 +67,13 @@ class SpeechForwarder : ObservableObject {
} }
guard let client = client else { return } guard let client = client else { return }
condition.lock()
client.close() client.close()
self.client = nil
condition.broadcast()
condition.unlock()
connected = false connected = false
} }
@ -64,25 +85,25 @@ class SpeechForwarder : ObservableObject {
// apps interface, so process the results on the apps // apps interface, so process the results on the apps
// main queue. // main queue.
OperationQueue.main.addOperation { OperationQueue.main.addOperation {
switch authStatus { switch authStatus {
case .authorized: case .authorized:
break break
case .denied: case .denied:
self.listening = false self.listening = false
break break
case .restricted: case .restricted:
self.listening = false self.listening = false
break break
case .notDetermined: case .notDetermined:
self.listening = false self.listening = false
break break
default: default:
self.listening = false self.listening = false
break break
} }
} }
} }
@ -111,16 +132,21 @@ class SpeechForwarder : ObservableObject {
if (!self.listening) { if (!self.listening) {
logger.debug("Stopped listening") logger.debug("Stopped listening")
audioEngine.stop()
recognitionRequest?.endAudio() recognitionRequest?.endAudio()
audioEngine.stop()
audioEngine.inputNode.removeTap(onBus: 0)
recognitionTask?.cancel() recognitionTask?.cancel()
audioEngine.inputNode.removeTap(onBus: 0);
audioEngine.inputNode.reset() self.recognitionRequest = nil
self.recognitionTask = nil
condition.lock()
self.listening = false
condition.broadcast()
condition.unlock()
switch (client.send(data: isListening())) { switch (client.send(data: isListening())) {
case .success: case .success:
break break
case .failure(let error): case .failure(let error):
self.listening = false
logger.error("Failed to send header: \(String(describing: error))") logger.error("Failed to send header: \(String(describing: error))")
} }
} }
@ -130,43 +156,109 @@ class SpeechForwarder : ObservableObject {
return pack("<hh", [LISTEN_STATE_MSG, listening ? 1 : 0]) return pack("<hh", [LISTEN_STATE_MSG, listening ? 1 : 0])
} }
private func send(latestText : String) { private func send() {
guard let client = client else { return } var stringLastSent = ""
var commonChars = self.textHeard.count var stringToSend = ""
var canSend = true
while true {
while (!canSend) {
logger.debug("Cannot send")
guard let client = client else {
logger.debug("Returning because client gone")
return
}
guard let byteArray = client.read(2, timeout: 1) else {
logger.debug("Did not read data")
continue
}
let data = Data(byteArray)
do {
let unpacked = try unpack("<h", data)
canSend = (unpacked[0] as? Int == LISTEN_SEND_MORE)
logger.debug("Updated canSend")
}
catch {
logger.debug("Unpack failed")
continue
}
}
logger.debug("Can send")
condition.lock()
while (stringLastSent == latestText) {
if (!self.listening) {
condition.unlock()
return
}
condition.wait()
if (!self.listening) {
condition.unlock()
return
}
guard client != nil else {
condition.unlock()
return
}
}
stringToSend = latestText
condition.unlock()
if send(latestText: stringToSend, lastSent: stringLastSent) {
stringLastSent = stringToSend
canSend = false
}
}
}
private func send(latestText : String, lastSent: String) -> Bool {
guard let client = client else { return false }
var commonChars = lastSent.count
while (commonChars > 0) { while (commonChars > 0) {
if (latestText.prefix(commonChars) == self.textHeard.prefix(commonChars)) { if (latestText.prefix(commonChars) == lastSent.prefix(commonChars)) {
break break
} }
commonChars -= 1 commonChars -= 1
} }
var stringToSend = "" var stringToSend = ""
if (commonChars < self.textHeard.count) { if (commonChars < lastSent.count) {
stringToSend = String(repeating: "\u{7f}", count: self.textHeard.count - commonChars) stringToSend = String(repeating: "\u{7f}", count: lastSent.count - commonChars)
} }
stringToSend.append(contentsOf: latestText.suffix(latestText.count - commonChars).replacingOccurrences(of: "\n", with: "\r")) stringToSend.append(contentsOf: latestText.suffix(latestText.count - commonChars).replacingOccurrences(of: "\n", with: "\r"))
if (stringToSend.count > 0) { if (stringToSend.count == 0) {
// TODO - Handle strings to send that are longer than 64K (doubt that would happen though) return false
let nsEnc = CFStringConvertEncodingToNSStringEncoding(CFStringEncoding(CFStringBuiltInEncodings.macRoman.rawValue)) }
let encoding = String.Encoding(rawValue: nsEnc) // String.Encoding
if let bytes = stringToSend.data(using: encoding) { // JSR_TODO - Handle strings to send that are longer than 64K (doubt that would happen though)
switch (client.send(data: pack("<hh", [LISTEN_TEXT_MSG, bytes.count]))) { let nsEnc = CFStringConvertEncodingToNSStringEncoding(CFStringEncoding(CFStringBuiltInEncodings.macRoman.rawValue))
case .success: let encoding = String.Encoding(rawValue: nsEnc) // String.Encoding
switch (client.send(data: bytes)) { if let bytes = stringToSend.data(using: encoding) {
case .success: switch (client.send(data: pack("<hh", [LISTEN_TEXT_MSG, bytes.count]))) {
self.textHeard = latestText case .success:
logger.debug("Sent text \"\(stringToSend)\"") switch (client.send(data: bytes)) {
break case .success:
case .failure(let error): logger.debug("Sent text \"\(stringToSend)\"")
self.listening = false break
logger.error("Failed to send text: \(String(describing: error))") case .failure(let error):
OperationQueue.main.addOperation {
if (self.listening) {
self.listen()
}
}
logger.error("Failed to send text: \(String(describing: error))")
return false
}
case .failure(let error):
OperationQueue.main.addOperation {
if (self.listening) {
self.listen()
} }
case .failure(let error): }
self.listening = false logger.error("Failed to send text: \(String(describing: error))")
logger.error("Failed to send text: \(String(describing: error))")
}
} }
} }
return true
} }
private func startRecording() throws { private func startRecording() throws {
@ -198,6 +290,13 @@ class SpeechForwarder : ObservableObject {
} }
self.textHeard = "" self.textHeard = ""
self.latestText = ""
self.sending = true
queue.addOperation {
self.send()
OperationQueue.main.addOperation { self.sending = false }
}
// Create a recognition task for the speech recognition session. // Create a recognition task for the speech recognition session.
// Keep a reference to the task so that it can be canceled. // Keep a reference to the task so that it can be canceled.
@ -206,9 +305,14 @@ class SpeechForwarder : ObservableObject {
if let result = result { if let result = result {
// Update the text view with the results. // Update the text view with the results.
self.send(latestText: result.bestTranscription.formattedString) self.condition.lock()
self.latestText = result.bestTranscription.formattedString
self.condition.broadcast()
self.condition.unlock()
OperationQueue.main.addOperation { self.textHeard = result.bestTranscription.formattedString }
isFinal = result.isFinal isFinal = result.isFinal
print("Text \(result.bestTranscription.formattedString)")
} }
if error != nil { if error != nil {
@ -216,20 +320,10 @@ class SpeechForwarder : ObservableObject {
} }
if error != nil || isFinal { if error != nil || isFinal {
// Stop recognizing speech if there is a problem. OperationQueue.main.addOperation {
self.audioEngine.stop() if (self.listening) {
inputNode.removeTap(onBus: 0) self.listen()
}
self.recognitionRequest = nil
self.recognitionTask = nil
self.listening = false
self.logger.debug("Stopped listening")
guard let client = self.client else { return }
switch (client.send(data: self.isListening())) {
case .success:
break
case .failure(let error):
self.logger.error("Failed to send header: \(String(describing: error))")
} }
} }
} }