From a2ca7a67df1f33d585b5d46a9bf5df307c2e3861 Mon Sep 17 00:00:00 2001
From: Shruti1229 <shrgupta01@gmail.com>
Date: Thu, 28 Dec 2023 00:28:05 +0530
Subject: [PATCH] Add Pre and post processor condition in Indic ASR

---
 backend/transcript/metadata.py  |  6 ++++--
 backend/transcript/utils/asr.py | 30 +++++++++++++++++++++---------
 2 files changed, 25 insertions(+), 11 deletions(-)

diff --git a/backend/transcript/metadata.py b/backend/transcript/metadata.py
index 91af3b61..9cb65937 100644
--- a/backend/transcript/metadata.py
+++ b/backend/transcript/metadata.py
@@ -2,7 +2,6 @@
 TRANSCRIPTION_LANGUAGE_CHOICES = [
     ("en", "English"),
     ("hi", "Hindi"),
-    ("as", "Assamese"),
     ("bn", "Bengali"),
     ("gu", "Gujarati"),
     ("kn", "Kannada"),
@@ -12,10 +11,11 @@
     ("pa", "Punjabi"),
     ("ta", "Tamil"),
     ("te", "Telugu"),
+    ("sa", "Sanskrit"),
+    ("ur": "Urdu"),
 ]
 
 TRANSCRIPTION_SUPPORTED_LANGUAGES = {
-    "Assamese": "as",
     "Bengali": "bn",
     "English": "en",
     "Gujarati": "gu",
@@ -25,6 +25,8 @@
     "Marathi": "mr",
     "Oriya": "or",
     "Punjabi": "pa",
+    "Sanskrit": "sa",
     "Tamil": "ta",
     "Telugu": "te",
+    "Urdu": "ur",
 }
diff --git a/backend/transcript/utils/asr.py b/backend/transcript/utils/asr.py
index 9e8c2289..51e118e4 100644
--- a/backend/transcript/utils/asr.py
+++ b/backend/transcript/utils/asr.py
@@ -55,15 +55,27 @@ def make_asr_api_call(url, lang, vad_level=3, chunk_size=10):
         else:
             return None
 
-        json_data = {
-            "config": {
-                "serviceId": service_id,
-                "language": {"sourceLanguage": lang},
-                "transcriptionFormat": {"value": "srt"},
-                "postProcessors": ["itn", "punctuation"],
-            },
-            "audio": [{"audioUri": url}],
-        }
+        if lang not in ["kn", "gu", "ur", "ml"]:
+            json_data = {
+                "config": {
+                    "serviceId": service_id,
+                    "language": {"sourceLanguage": lang},
+                    "transcriptionFormat": {"value": "srt"},
+                    "postProcessors": ["itn", "punctuation"],
+                    "preProcessors": ["vad"],
+                },
+                "audio": [{"audioUri": url}],
+            }
+        else:
+            json_data = {
+                "config": {
+                    "serviceId": service_id,
+                    "language": {"sourceLanguage": lang},
+                    "transcriptionFormat": {"value": "srt"},
+                    "preProcessors": ["vad"],
+                },
+                "audio": [{"audioUri": url}],
+            }
         logging.info("Sending request to indic model.")
         try:
             response = requests.post(