Normalize loose STT transcripts before routing

OpenJibo/src/Jibo.Cloud/dotnet/src/Jibo.Cloud.Application/Services/SyntheticBufferedAudioSttStrategy.cs

@@ -20,7 +20,7 @@ public sealed class SyntheticBufferedAudioSttStrategy : ISttStrategy
 
         return Task.FromResult(new SttResult
         {
-            Text = transcriptHint.Trim(),
+            Text = NormalizeLooseTranscript(transcriptHint),
             Provider = Name,
             Confidence = 0.75f,
             Locale = turn.Locale,
@@ -51,4 +51,16 @@ public sealed class SyntheticBufferedAudioSttStrategy : ISttStrategy
             ? transcriptHint?.ToString()
             : null;
     }
+
+    private static string NormalizeLooseTranscript(string? value)
+    {
+        if (string.IsNullOrWhiteSpace(value)) return string.Empty;
+
+        var lowered = value.Trim().ToLowerInvariant();
+        lowered = System.Text.RegularExpressions.Regex.Replace(lowered, @"[^\p{L}\p{N}\s']+", " ",
+            System.Text.RegularExpressions.RegexOptions.CultureInvariant | System.Text.RegularExpressions.RegexOptions.Compiled);
+        lowered = System.Text.RegularExpressions.Regex.Replace(lowered, @"\s+"," ",
+            System.Text.RegularExpressions.RegexOptions.CultureInvariant | System.Text.RegularExpressions.RegexOptions.Compiled);
+        return lowered.Trim();
+    }
 }

OpenJibo/src/Jibo.Cloud/dotnet/src/Jibo.Cloud.Infrastructure/Audio/AudioTranscriptNormalizer.cs

@@ -0,0 +1,24 @@
+using System.Text.RegularExpressions;
+
+namespace Jibo.Cloud.Infrastructure.Audio;
+
+internal static class AudioTranscriptNormalizer
+{
+    private static readonly Regex PunctuationToSpaceRegex = new(
+        @"[^\p{L}\p{N}\s']+",
+        RegexOptions.CultureInvariant | RegexOptions.Compiled);
+
+    private static readonly Regex WhitespaceRegex = new(
+        @"\s+",
+        RegexOptions.CultureInvariant | RegexOptions.Compiled);
+
+    public static string NormalizeLooseTranscript(string? value)
+    {
+        if (string.IsNullOrWhiteSpace(value)) return string.Empty;
+
+        return WhitespaceRegex.Replace(
+                PunctuationToSpaceRegex.Replace(value.Trim().ToLowerInvariant(), " "),
+                " ")
+            .Trim();
+    }
+}

OpenJibo/src/Jibo.Cloud/dotnet/src/Jibo.Cloud.Infrastructure/Audio/LocalWhisperCppBufferedAudioSttStrategy.cs

@@ -52,6 +52,7 @@ public sealed class LocalWhisperCppBufferedAudioSttStrategy(
                 cancellationToken);
 
             var transcript = ExtractTranscript(whisperResult.StdOut);
+            transcript = AudioTranscriptNormalizer.NormalizeLooseTranscript(transcript);
             if (string.IsNullOrWhiteSpace(transcript))
                 throw new InvalidOperationException("whisper.cpp returned no transcript for the buffered audio turn.");
 

OpenJibo/tests/Jibo.Cloud.Tests/WebSockets/LocalWhisperCppBufferedAudioSttStrategyTests.cs

@@ -123,6 +123,48 @@ public sealed class LocalWhisperCppBufferedAudioSttStrategyTests
         }
     }
 
+    [Fact]
+    public async Task TranscribeAsync_NormalizesLoosePunctuationFromWhisperOutput()
+    {
+        var tempDirectory = Path.Combine(Path.GetTempPath(), $"openjibo-stt-test-{Guid.NewGuid():N}");
+        Directory.CreateDirectory(tempDirectory);
+
+        try
+        {
+            var runner = new FakeExternalProcessRunner("[00:00:00.000 --> 00:00:01.000] - Thank you. - Yes.");
+            var strategy = new LocalWhisperCppBufferedAudioSttStrategy(
+                new BufferedAudioSttOptions
+                {
+                    EnableLocalWhisperCpp = true,
+                    FfmpegPath = "ffmpeg",
+                    WhisperCliPath = "whisper-cli",
+                    WhisperModelPath = "model.bin",
+                    TempDirectory = tempDirectory
+                },
+                runner);
+
+            var turn = new TurnContext
+            {
+                TurnId = "turn-local-stt-punctuation",
+                Locale = "en-US",
+                Attributes = new Dictionary<string, object?>
+                {
+                    ["bufferedAudioBytes"] = 47,
+                    ["bufferedAudioFrames"] = new[] { BuildMinimalOggPage() }
+                }
+            };
+
+            var result = await strategy.TranscribeAsync(turn);
+
+            Assert.Equal("thank you yes", result.Text);
+            Assert.Equal("local-whispercpp-buffered-audio", result.Provider);
+        }
+        finally
+        {
+            if (Directory.Exists(tempDirectory)) Directory.Delete(tempDirectory, true);
+        }
+    }
+
     private static byte[] BuildMinimalOggPage()
     {
         return
@@ -148,7 +190,8 @@ public sealed class LocalWhisperCppBufferedAudioSttStrategyTests
         return page;
     }
 
-    private sealed class FakeExternalProcessRunner : IExternalProcessRunner
+    private sealed class FakeExternalProcessRunner(string whisperStdOut = "[00:00:00.000 --> 00:00:01.000] tell me a joke")
+        : IExternalProcessRunner
     {
         public List<(string FileName, IReadOnlyList<string> Arguments)> Calls { get; } = [];
 
@@ -158,8 +201,7 @@ public sealed class LocalWhisperCppBufferedAudioSttStrategyTests
             Calls.Add((fileName, arguments));
 
             if (!string.Equals(fileName, "ffmpeg", StringComparison.OrdinalIgnoreCase))
-                return Task.FromResult(new ExternalProcessResult(0, "[00:00:00.000 --> 00:00:01.000] tell me a joke",
+                return Task.FromResult(new ExternalProcessResult(0, whisperStdOut, string.Empty));
-                    string.Empty));
 
             var outputPath = arguments[^1];
             File.WriteAllBytes(outputPath, "RIFF"u8);

OpenJibo/tests/Jibo.Cloud.Tests/WebSockets/SyntheticBufferedAudioSttStrategyTests.cs

@@ -0,0 +1,25 @@
+using Jibo.Cloud.Application.Services;
+using Jibo.Runtime.Abstractions;
+
+namespace Jibo.Cloud.Tests.WebSockets;
+
+public sealed class SyntheticBufferedAudioSttStrategyTests
+{
+    [Fact]
+    public async Task TranscribeAsync_NormalizesLoosePunctuationInTranscriptHint()
+    {
+        var strategy = new SyntheticBufferedAudioSttStrategy();
+        var result = await strategy.TranscribeAsync(new TurnContext
+        {
+            Attributes = new Dictionary<string, object?>
+            {
+                ["bufferedAudioBytes"] = 42,
+                ["audioTranscriptHint"] = "- Thank you. - Yes."
+            }
+        });
+
+        Assert.Equal("thank you yes", result.Text);
+        Assert.Equal("synthetic-buffered-audio", result.Provider);
+        Assert.Equal(0.75f, result.Confidence);
+    }
+}