From a6e5838d788ea333e2c7eb6d100fb0d754d83a02 Mon Sep 17 00:00:00 2001
From: tomkarho <tommi@tomkarho.com>
Date: Wed, 8 May 2024 13:14:57 +0300
Subject: [PATCH] Handle audio up to 30 seconds

---
 AzureAi.Transcriber/.gitignore                |  3 +-
 .../AzureAi.Transcriber.csproj                |  4 ++
 AzureAi.Transcriber/Program.cs                |  7 +++
 .../Services/TranscribeService.cs             | 43 +++++++++++++++++++
 4 files changed, 56 insertions(+), 1 deletion(-)
 create mode 100644 AzureAi.Transcriber/Services/TranscribeService.cs

diff --git a/AzureAi.Transcriber/.gitignore b/AzureAi.Transcriber/.gitignore
index 8d4a6c0..d9a8393 100644
--- a/AzureAi.Transcriber/.gitignore
+++ b/AzureAi.Transcriber/.gitignore
@@ -1,2 +1,3 @@
 bin
-obj
\ No newline at end of file
+obj
+Data/*
\ No newline at end of file
diff --git a/AzureAi.Transcriber/AzureAi.Transcriber.csproj b/AzureAi.Transcriber/AzureAi.Transcriber.csproj
index 13d8127..d250f2e 100644
--- a/AzureAi.Transcriber/AzureAi.Transcriber.csproj
+++ b/AzureAi.Transcriber/AzureAi.Transcriber.csproj
@@ -6,4 +6,8 @@
     <ImplicitUsings>enable</ImplicitUsings>
   </PropertyGroup>
 
+  <ItemGroup>
+    <PackageReference Include="Microsoft.CognitiveServices.Speech" Version="1.37.0" />
+  </ItemGroup>
+
 </Project>
diff --git a/AzureAi.Transcriber/Program.cs b/AzureAi.Transcriber/Program.cs
index 2c67dd3..9c53f9c 100644
--- a/AzureAi.Transcriber/Program.cs
+++ b/AzureAi.Transcriber/Program.cs
@@ -1,4 +1,5 @@
 using AzureAi.Transcriber.Components;
+using AzureAi.Transcriber.Services;
 
 var builder = WebApplication.CreateBuilder(args);
 
@@ -6,8 +7,14 @@ var builder = WebApplication.CreateBuilder(args);
 builder.Services.AddRazorComponents()
     .AddInteractiveServerComponents();
 
+builder.Services.AddSingleton<IFileService, FileService>();
+builder.Services.AddSingleton<ITranscribeService, TranscribeService>();
+
 var app = builder.Build();
 
+// Warmup
+app.Services.GetRequiredService<ITranscribeService>();
+
 // Configure the HTTP request pipeline.
 if (!app.Environment.IsDevelopment())
 {
diff --git a/AzureAi.Transcriber/Services/TranscribeService.cs b/AzureAi.Transcriber/Services/TranscribeService.cs
new file mode 100644
index 0000000..07c829d
--- /dev/null
+++ b/AzureAi.Transcriber/Services/TranscribeService.cs
@@ -0,0 +1,43 @@
+using Microsoft.CognitiveServices.Speech;
+using Microsoft.CognitiveServices.Speech.Audio;
+
+namespace AzureAi.Transcriber.Services;
+
+public interface ITranscribeService
+{
+    Task<SpeechRecognitionResult> Transcribe(string filePath);
+}
+
+public class TranscribeService: ITranscribeService
+{
+    private static string _speechKey = Environment.GetEnvironmentVariable("SPEECH_KEY") ?? string.Empty;
+    private static string _speechRegion = Environment.GetEnvironmentVariable("SPEECH_REGION") ?? string.Empty;
+
+    private ILogger<TranscribeService> Logger;
+    
+    public TranscribeService(ILogger<TranscribeService> logger)
+    {
+        Logger = logger;
+        
+        if(string.IsNullOrWhiteSpace(_speechKey) || string.IsNullOrWhiteSpace(_speechRegion))
+        {
+            throw new InvalidOperationException("Speech key and region must be set in environment variables.");
+        }
+    }
+    
+    public async Task<SpeechRecognitionResult> Transcribe(string filePath)
+    {
+        Logger.LogInformation("Transcribing {filePath}", filePath);
+        await Task.Delay(3000);
+        
+        var speechConfig = SpeechConfig.FromSubscription(_speechKey, _speechRegion);
+        speechConfig.SpeechRecognitionLanguage = "en-US";
+        
+        using var audioConfig = AudioConfig.FromWavFileInput(filePath);
+        using var recognizer = new SpeechRecognizer(speechConfig, audioConfig);
+        
+        var result = await recognizer.RecognizeOnceAsync();
+
+        return result;
+    }
+}
\ No newline at end of file