v0.7: pre-synth buffering, session IDs, segment timeout playback

Robovoice.Stt.Dhcp/DhcpSttSource.cs

@@ -19,6 +19,7 @@ public sealed class DhcpSttSource : ISttSource
     private Task? _nopTask;
     private EndPoint _broadcastEp = new IPEndPoint(IPAddress.Broadcast, DhcpServerPort);
     private uint _nonce;
+    private uint _session;
     private bool _disposed;
 
     public string InterfaceIp { get; set; } = string.Empty;
@@ -73,7 +74,9 @@ public sealed class DhcpSttSource : ISttSource
         if (_cts is null)
             return;
 
+        _session++;
         _nonce = 0;
+        Log?.Invoke($"STT: session {_session} started");
         SendNop();
         _nopTask = NopLoopAsync(_cts.Token);
     }
@@ -81,7 +84,7 @@ public sealed class DhcpSttSource : ISttSource
     public void SendOff()
     {
         StopNop();
-        SendControl("HKMSTR:OFF");
+        SendControl($"HKMSTR:OFF {_session} {_nonce}");
     }
 
     private void StopNop()
@@ -107,7 +110,7 @@ public sealed class DhcpSttSource : ISttSource
     private void SendNop()
     {
         _nonce++;
-        SendControl($"HKMSTR {_nonce}");
+        SendControl($"HKMSTR {_session} {_nonce}");
     }
 
     private void SendControl(string message)
@@ -158,25 +161,67 @@ public sealed class DhcpSttSource : ISttSource
             if (!text.StartsWith(Magic))
                 continue;
 
-            if (text.StartsWith("HKMSTR:P "))
+            TranscriptMessage? message = ParseReply(text);
+            if (message is null)
+                continue;
+
+            TranscriptReceived?.Invoke(this, new TranscriptEventArgs
             {
-                string transcript = text["HKMSTR:P ".Length..];
-                TranscriptReceived?.Invoke(this, new TranscriptEventArgs
-                {
-                    Message = new TranscriptMessage(TranscriptType.Partial, transcript),
-                });
-            }
-            else if (text.StartsWith("HKMSTR:F "))
-            {
-                string transcript = text["HKMSTR:F ".Length..];
-                TranscriptReceived?.Invoke(this, new TranscriptEventArgs
-                {
-                    Message = new TranscriptMessage(TranscriptType.Final, transcript),
-                });
-            }
+                Message = message,
+            });
         }
     }
 
+    private TranscriptMessage? ParseReply(string text)
+    {
+        // Format: HKMSTR:P <session> <text>  or  HKMSTR:F <session> <text>
+        // <text> may be empty.
+        string prefix;
+        TranscriptType type;
+
+        if (text.StartsWith("HKMSTR:P "))
+        {
+            prefix = "HKMSTR:P ";
+            type = TranscriptType.Partial;
+        }
+        else if (text.StartsWith("HKMSTR:F "))
+        {
+            prefix = "HKMSTR:F ";
+            type = TranscriptType.Final;
+        }
+        else
+        {
+            return null;
+        }
+
+        string rest = text[prefix.Length..];
+
+        int spaceIndex = rest.IndexOf(' ');
+        if (spaceIndex < 0)
+        {
+            if (uint.TryParse(rest, out uint sessionOnly))
+            {
+                if (sessionOnly != _session)
+                    return null;
+                return new TranscriptMessage(type, string.Empty);
+            }
+            return null;
+        }
+
+        string sessionStr = rest[..spaceIndex];
+        if (!uint.TryParse(sessionStr, out uint session))
+            return null;
+
+        if (session != _session)
+        {
+            Log?.Invoke($"STT: dropping stale reply (session {session} != current {_session})");
+            return null;
+        }
+
+        string transcript = rest[(spaceIndex + 1)..];
+        return new TranscriptMessage(type, transcript);
+    }
+
     public static List<(string Ip, string Name)> GetAvailableInterfaces()
     {
         var result = new List<(string, string)>();