Update app.py

app.py

@@ -7,6 +7,11 @@ import os
 
 # Assuming other necessary imports and setup are already done
 
+auth_token = os.environ.get("HF_TOKEN")
+target_lang_options = {"English": "eng", "Luganda": "lug", "Acholi": "ach", "Runyankole": "nyn", "Lugbara": "lgg"}
+languages = list(target_lang_options.keys())
+
+
 # Helper function to format and group word timestamps
 def format_and_group_timestamps(chunks, interval=5.0):
     grouped = {}
@@ -35,11 +40,11 @@ def transcribe_audio(input_file, language, chunk_length_s=10, stride_length_s=(4
     if target_lang_code == "eng":
         model_id = "facebook/mms-1b-all"
     else:
-        model_id = "custom_model_id_for_other_languages"  # Placeholder for actual model IDs
+        model_id = "Sunbird/sunbird-mms"
                 
-    auth_token = os.environ.get("HF_TOKEN")
     pipe = pipeline(model=model_id, device=device, token=auth_token)  
-    # Assuming necessary setup for tokenizer and loading adapter
+    pipe.tokenizer.set_target_lang(target_lang_code)
+    pipe.model.load_adapter(target_lang_code)
 
     output = pipe(input_file, chunk_length_s=chunk_length_s, stride_length_s=stride_length_s, return_timestamps=return_timestamps)
     formatted_output = format_and_group_timestamps(output['chunks'])