support multi-lingual ASR models (#723)

## Summary

Update live audio transcription samples to use the correct Foundry Local
catalog alias for the multilingual Nemotron ASR model, and strip
model-emitted language tags in the SDK so all consumers receive clean
transcription text.

## Changes

### Samples (all 5 languages)
- Updated the commented multilingual model alias from
`Nemotron-3.5-ASR-Streaming-Multilingual-0.6b-onnx-int4` to
`nvidia-nemotron-3.5-asr-streaming-multilingual-0.6b` to match the FL
catalog naming convention (mirrors the HuggingFace repo name)
- Affected files:
  - `samples/cs/live-audio-transcription/Program.cs`
  - `samples/js/live-audio-transcription/app.js`
  - `samples/python/live-audio-transcription/src/app.py`
  - `samples/cpp/live-audio-transcription/main.cpp`
  - `samples/rust/live-audio-transcription/src/main.rs`

### SDK (`sdk/cs/src/OpenAI/LiveAudioTypes.cs`)
- Strip language identification tags (e.g. `<zh-CN>`, `<en-US>`) from
transcription text in `LiveAudioTranscriptionResponse.FromJson()`
- Multilingual Nemotron models emit these tags at segment boundaries as
language identifiers — following industry convention (Whisper, Azure
Speech, Google STT), these should be returned as metadata rather than
inline text
- All SDK consumers now receive clean text without needing client-side
post-processing

## Testing

Verified end-to-end with
`nvidia-nemotron-3.5-asr-streaming-multilingual-0.6b` (INT4) model:
- English speech → clean English transcription
- Chinese speech (`zh-CN`) → clean Chinese transcription
- Mixed language switching works correctly
- Language tags stripped at the SDK layer before reaching consumers

Author: rui-ren

samples/cpp/live-audio-transcription/main.cpp

@@ -126,9 +126,13 @@ int main(int argc, char* argv[]) {
         manager.DownloadAndRegisterEps(nullptr, isCancellationRequested);
 
         auto& catalog = manager.GetCatalog();
-        auto* model = catalog.GetModel("nemotron-speech-streaming-en-0.6b");
+        // English-only:
+        const char* modelAlias = "nemotron-speech-streaming-en-0.6b";
+        // Multi-lingual (supports 30+ languages including auto-detect):
+        // const char* modelAlias = "nvidia-nemotron-3.5-asr-streaming-multilingual-0.6b";
+        auto* model = catalog.GetModel(modelAlias);
         if (!model) {
-            throw std::runtime_error("Model \"nemotron-speech-streaming-en-0.6b\" not found in catalog");
+            throw std::runtime_error(std::string("Model \"") + modelAlias + "\" not found in catalog");
         }
 
         std::cout << "Downloading model (if needed)..." << std::endl;
@@ -151,7 +155,11 @@ int main(int argc, char* argv[]) {
         session->Settings().sample_rate = 16000;
         session->Settings().channels = 1;
         session->Settings().bits_per_sample = 16;
-        session->Settings().language = "en";
+        session->Settings().language = "en";                  // English (default)
+        // Multi-lingual examples:
+        // session->Settings().language = "de";     // German
+        // session->Settings().language = "zh-CN";  // Chinese (Simplified)
+        // session->Settings().language = "auto";   // Auto-detect language
         session->Start();
         std::cout << "Session started" << std::endl;
 

samples/cs/live-audio-transcription/Program.cs

@@ -24,7 +24,12 @@
 
 var catalog = await mgr.GetCatalogAsync();
 
-var model = await catalog.GetModelAsync("nemotron-speech-streaming-en-0.6b") ?? throw new Exception("Model \"nemotron-speech-streaming-en-0.6b\" not found in catalog");
+// English-only:
+var modelAlias = "nemotron-speech-streaming-en-0.6b";
+// Multi-lingual (supports 30+ languages including auto-detect):
+// var modelAlias = "nvidia-nemotron-3.5-asr-streaming-multilingual-0.6b";
+
+var model = await catalog.GetModelAsync(modelAlias) ?? throw new Exception($"Model \"{modelAlias}\" not found in catalog");
 
 await model.DownloadAsync(progress =>
 {
@@ -43,7 +48,11 @@ await model.DownloadAsync(progress =>
 var session = audioClient.CreateLiveTranscriptionSession();
 session.Settings.SampleRate = 16000;  // Default is 16000; shown here to match the NAudio WaveFormat below
 session.Settings.Channels = 1;
-session.Settings.Language = "en";
+session.Settings.Language = "en";                  // English (default)
+// Multi-lingual examples:
+// session.Settings.Language = "de";     // German
+// session.Settings.Language = "zh-CN";  // Chinese (Simplified)
+// session.Settings.Language = "auto";   // Auto-detect language
 
 await session.StartAsync();
 Console.WriteLine("       Session started");

samples/js/live-audio-transcription/app.js

@@ -21,7 +21,10 @@ const manager = FoundryLocalManager.create({
 console.log('✓ SDK initialized');
 
 // Get and load the nemotron model
+// English-only:
 const modelAlias = 'nemotron-speech-streaming-en-0.6b';
+// Multi-lingual (supports 30+ languages including auto-detect):
+// const modelAlias = 'nvidia-nemotron-3.5-asr-streaming-multilingual-0.6b';
 let model = await manager.catalog.getModel(modelAlias);
 if (!model) {
     console.error(`ERROR: Model "${modelAlias}" not found in catalog.`);
@@ -46,7 +49,11 @@ const session = audioClient.createLiveTranscriptionSession();
 session.settings.sampleRate = 16000;  // Default is 16000; shown here for clarity
 session.settings.channels = 1;
 session.settings.bitsPerSample = 16;
-session.settings.language = 'en';
+session.settings.language = 'en';                  // English (default)
+// Multi-lingual examples:
+// session.settings.language = 'de';     // German
+// session.settings.language = 'zh-CN';  // Chinese (Simplified)
+// session.settings.language = 'auto';   // Auto-detect language
 
 console.log('Starting streaming session...');
 await session.start();

samples/python/live-audio-transcription/src/app.py

@@ -29,9 +29,13 @@
 
 manager.download_and_register_eps()
 
-model = manager.catalog.get_model("nemotron-speech-streaming-en-0.6b")
+# English-only:
+model_alias = "nemotron-speech-streaming-en-0.6b"
+# Multi-lingual (supports 30+ languages including auto-detect):
+# model_alias = "nvidia-nemotron-3.5-asr-streaming-multilingual-0.6b"
+model = manager.catalog.get_model(model_alias)
 if model is None:
-    raise RuntimeError('Model "nemotron-speech-streaming-en-0.6b" not found in catalog')
+    raise RuntimeError(f'Model "{model_alias}" not found in catalog')
 
 model.download(
     lambda progress: print(f"\rDownloading model: {progress:.2f}%", end="", flush=True)
@@ -45,7 +49,11 @@
 session = audio_client.create_live_transcription_session()
 session.settings.sample_rate = 16000
 session.settings.channels = 1
-session.settings.language = "en"
+session.settings.language = "en"                  # English (default)
+# Multi-lingual examples:
+# session.settings.language = "de"     # German
+# session.settings.language = "zh-CN"  # Chinese (Simplified)
+# session.settings.language = "auto"   # Auto-detect language
 
 session.start()
 print("✓ Session started")

samples/rust/live-audio-transcription/src/main.rs

@@ -15,7 +15,10 @@ use cpal::traits::{DeviceTrait, HostTrait, StreamTrait};
 use foundry_local_sdk::{FoundryLocalConfig, FoundryLocalManager, LiveAudioTranscriptionSession};
 use tokio_stream::StreamExt;
 
+// English-only:
 const ALIAS: &str = "nemotron-speech-streaming-en-0.6b";
+// Multi-lingual (supports 30+ languages including auto-detect):
+// const ALIAS: &str = "nvidia-nemotron-3.5-asr-streaming-multilingual-0.6b";
 
 // Global flag for Ctrl+C graceful shutdown (mirrors JS process.on('SIGINT'))
 static RUNNING: AtomicBool = AtomicBool::new(true);
@@ -57,7 +60,12 @@ async fn main() -> Result<(), Box<dyn std::error::Error>> {
     println!("✓ Model loaded\n");
 
     let audio_client = model.create_audio_client();
-    let session = Arc::new(audio_client.create_live_transcription_session());
+    let mut session = audio_client.create_live_transcription_session();
+    session.settings.language = Some("en".into());    // English (default)
+    // session.settings.language = Some("de".into());    // German
+    // session.settings.language = Some("zh-CN".into()); // Chinese (Simplified)
+    // session.settings.language = Some("auto".into());  // Auto-detect language
+    let session = Arc::new(session);
     session.start(None).await?;
     println!("✓ Session started\n");
 

sdk/cs/src/OpenAI/LiveAudioTypes.cs

@@ -32,6 +32,10 @@ public class LiveAudioTranscriptionResponse : ConversationItem
     [JsonPropertyName("end_time")]
     public double? EndTime { get; init; }
 
+    /// <summary>BCP-47 language code detected by the model (e.g. "zh-CN", "en-US"), or null if not detected.</summary>
+    [JsonPropertyName("language")]
+    public string? Language { get; init; }
+
     internal static LiveAudioTranscriptionResponse FromJson(string json)
     {
         var raw = JsonSerializer.Deserialize(json,
@@ -43,6 +47,7 @@ internal static LiveAudioTranscriptionResponse FromJson(string json)
             IsFinal = raw.IsFinal,
             StartTime = raw.StartTime,
             EndTime = raw.EndTime,
+            Language = raw.Language,
             Content =
             [
                 new ContentPart
@@ -72,6 +77,9 @@ internal record LiveAudioTranscriptionRaw
 
     [JsonPropertyName("end_time")]
     public double? EndTime { get; init; }
+
+    [JsonPropertyName("language")]
+    public string? Language { get; init; }
 }
 
 internal record CoreErrorResponse