Hindi Support

proxies.txt

@@ -0,0 +1,300 @@
+158.255.77.168:80
+37.120.172.84:80
+50.223.246.237:80
+50.174.7.159:80
+50.207.199.87:80
+32.223.6.94:80
+89.58.53.205:80
+50.207.199.80:80
+50.207.199.83:80
+50.174.7.153:80
+165.232.129.150:80
+50.239.72.18:80
+50.175.212.66:80
+50.217.226.47:80
+50.239.72.16:80
+50.221.74.130:80
+190.58.248.86:80
+50.207.199.82:80
+50.174.7.152:80
+50.122.86.118:80
+66.191.31.158:80
+89.58.45.248:80
+89.58.28.110:80
+202.61.199.166:80
+89.58.55.106:80
+23.247.136.254:80
+3.110.60.103:80
+89.58.52.160:80
+54.174.151.201:80
+37.19.208.90:8080
+50.217.226.43:80
+50.239.72.17:80
+50.174.7.157:80
+50.217.226.44:80
+50.174.7.158:80
+50.217.226.42:80
+50.174.7.155:80
+50.174.7.162:80
+213.143.113.82:80
+50.221.230.186:80
+50.217.226.41:80
+50.207.199.86:80
+0.0.0.0:80
+211.128.96.206:80
+68.185.57.66:80
+50.174.7.156:80
+50.207.199.81:80
+127.0.0.7:80
+99.79.124.70:80
+109.94.208.128:8888
+40.76.69.94:8080
+162.223.90.150:80
+66.201.7.151:3128
+3.26.71.155:80
+47.251.122.81:8888
+50.202.75.26:80
+50.239.72.19:80
+50.175.212.74:80
+188.68.52.244:80
+193.30.122.197:80
+89.58.8.250:80
+95.216.148.196:80
+23.247.136.248:80
+173.212.208.108:2082
+45.140.143.77:18080
+200.250.131.218:80
+89.58.55.33:80
+47.88.59.79:82
+43.130.47.134:8080
+119.156.195.173:3128
+159.69.57.20:8880
+50.231.104.58:80
+8.217.124.178:49440
+213.148.10.80:3128
+103.227.187.1:6080
+45.123.142.11:8181
+38.159.229.97:999
+186.96.160.202:999
+139.59.1.14:80
+129.226.155.235:8080
+219.65.73.81:80
+66.29.154.103:3128
+50.217.226.40:80
+81.169.213.169:8888
+41.59.90.171:80
+84.39.112.144:3128
+47.56.110.204:8989
+128.140.113.110:3128
+47.238.149.53:8888
+159.65.245.255:80
+198.49.68.80:80
+88.99.209.189:1234
+159.203.61.169:3128
+34.143.143.61:7777
+134.209.29.120:80
+103.249.120.167:80
+44.220.205.79:8080
+103.75.119.185:80
+198.199.86.11:8080
+51.159.98.163:8089
+70.153.208.157:8080
+57.128.37.47:3128
+181.119.86.186:999
+47.91.115.179:8081
+149.129.255.179:80
+162.243.149.86:31028
+172.188.122.92:80
+172.167.161.8:8080
+171.248.200.229:10001
+103.214.102.172:8083
+103.125.174.5:8080
+103.48.71.162:82
+186.190.225.152:999
+171.22.8.192:80
+47.251.87.199:1036
+5.45.126.128:8080
+158.255.77.169:80
+72.10.164.178:11517
+104.223.103.218:3428
+137.184.174.32:4857
+200.174.198.86:8888
+41.59.90.175:80
+65.38.97.146:3128
+155.94.128.59:10809
+198.74.51.79:8888
+103.82.134.199:8888
+161.35.70.249:8080
+47.236.224.32:8080
+78.108.38.82:8118
+87.248.129.26:80
+213.149.184.6:23128
+178.208.168.81:8118
+87.106.103.177:3128
+156.67.28.21:8080
+85.215.64.49:80
+192.73.244.36:80
+144.126.216.57:80
+23.82.137.157:80
+143.42.66.91:80
+185.94.35.26:8085
+193.233.211.141:8085
+146.19.44.108:8085
+193.233.231.194:8085
+104.239.13.32:6661
+23.27.75.71:6151
+31.57.82.108:6689
+185.226.204.95:5648
+64.49.37.158:8085
+166.88.58.222:5947
+193.233.140.72:8085
+198.105.122.193:6766
+185.216.106.213:6290
+92.113.241.107:6192
+185.68.152.228:8085
+162.220.247.170:6765
+185.226.207.17:5566
+173.239.219.223:6132
+83.97.119.200:8085
+31.58.30.194:6776
+199.96.166.148:8085
+64.49.38.215:8085
+193.233.82.203:8085
+142.111.113.15:6376
+185.94.32.47:8085
+185.226.207.166:5715
+193.233.229.204:8085
+64.49.36.46:8085
+91.243.91.132:8085
+185.68.152.178:8085
+140.235.2.65:8085
+91.243.90.205:8085
+206.206.69.5:6269
+107.181.154.141:5819
+89.249.197.127:6713
+185.77.223.190:8085
+88.218.46.93:8085
+86.38.154.231:5874
+140.235.3.77:8085
+45.10.165.133:8085
+140.235.2.33:8085
+193.233.230.91:8085
+193.233.143.130:8085
+185.226.204.189:5742
+140.235.0.67:8085
+104.143.226.241:5844
+45.66.209.43:8085
+45.43.70.178:6465
+212.119.41.227:8085
+84.46.204.43:6346
+86.38.236.54:6338
+194.110.150.96:8085
+141.98.85.122:8085
+45.43.64.110:6368
+185.88.101.15:8085
+198.105.122.86:6659
+89.116.78.214:5825
+212.18.113.217:8085
+45.132.185.241:8085
+199.96.165.117:8085
+91.243.91.201:8085
+5.183.253.227:8085
+193.233.210.24:8085
+91.243.94.11:8085
+104.143.224.11:5872
+45.145.129.183:8085
+142.111.58.110:6688
+104.239.13.130:6759
+166.88.58.147:5872
+89.116.78.197:5808
+185.88.101.237:8085
+193.233.83.162:8085
+67.227.112.62:6102
+64.49.36.235:8085
+193.202.16.125:8085
+193.233.143.85:8085
+199.96.164.171:8085
+193.93.192.235:8085
+91.243.89.75:8085
+198.105.122.50:6623
+185.226.204.70:5623
+146.103.3.72:7125
+142.147.131.230:6130
+85.239.56.246:8085
+45.80.107.31:8085
+91.243.90.155:8085
+140.235.2.45:8085
+62.204.35.110:8085
+176.126.104.184:8085
+193.233.231.67:8085
+104.239.13.2:6631
+45.145.128.89:8085
+193.233.140.247:8085
+91.243.95.66:8085
+88.218.46.235:8085
+85.239.58.173:8085
+185.68.155.103:8085
+193.163.207.13:8085
+193.233.138.76:8085
+193.233.231.224:8085
+31.57.41.125:5701
+89.116.78.80:5691
+91.243.90.44:8085
+45.80.107.196:8085
+45.145.129.47:8085
+64.49.37.221:8085
+104.143.226.20:5623
+193.233.248.99:8085
+45.43.81.114:5761
+193.233.143.227:8085
+104.239.13.55:6684
+86.38.26.123:6288
+181.41.218.64:5650
+146.103.3.121:7174
+31.57.41.93:5669
+91.243.89.113:8085
+193.233.137.57:8085
+199.96.165.241:8085
+193.93.193.130:8085
+193.93.195.130:8085
+84.46.204.223:6526
+167.253.49.95:8085
+199.96.165.212:8085
+193.233.143.60:8085
+146.103.3.131:7184
+45.43.70.36:6323
+170.106.184.175:10212
+62.204.35.227:8085
+31.57.82.187:6768
+140.235.1.169:8085
+193.233.82.20:8085
+193.233.210.251:8085
+154.30.252.153:5284
+45.43.81.153:5800
+193.202.16.168:8085
+217.69.121.57:5722
+146.185.202.177:8085
+193.233.228.182:8085
+193.233.143.18:8085
+104.239.13.22:6651
+92.113.241.55:6140
+167.253.48.251:8085
+199.96.167.253:8085
+64.49.36.201:8085
+45.159.23.254:8085
+31.57.82.213:6794
+198.105.111.242:6920
+198.46.241.80:6615
+146.185.203.80:8085
+185.226.204.51:5604
+104.238.10.77:6023
+199.96.165.131:8085
+193.233.231.186:8085
+69.30.217.114:17016
+198.46.241.96:6631
+140.235.1.233:8085
+31.57.41.42:5618
+107.181.154.72:5750
+31.57.41.193:5769
+64.49.38.107:8085
+146.103.3.167:7220

speech_utils.py

@@ -10,7 +10,26 @@ import asyncio # Import asyncio for await
 # Configure logging
 logging.basicConfig(level=logging.INFO)
 logger = logging.getLogger(__name__)
-
+import random
+
+def get_random_proxy():
+    proxies_path = os.path.join(os.path.dirname(__file__), "proxies.txt")
+    with open(proxies_path, "r") as f:
+        proxies = [line.strip() for line in f if line.strip()]
+    if not proxies:
+        return None
+    return random.choice(proxies)
+
+def build_translator_with_proxy():
+    proxy = get_random_proxy()
+    if proxy:
+        proxy_url = f"http://{proxy}"
+        return Translator(
+            service_urls=['translate.googleapis.com'],
+            proxies={"http": proxy_url, "https": proxy_url}
+        )
+    else:
+        return Translator(service_urls=['translate.googleapis.com'])
 # Define supported languages (using short codes consistent with Whisper/googletrans)
 # Note: googletrans uses short codes like 'en', 'hi'. Whisper also detects these.
 SUPPORTED_LANGUAGES = {
@@ -46,7 +65,7 @@ except Exception as e:
     model = None
 
 # Initialize the translator
-translator = Translator()
+translator = build_translator_with_proxy()
 
 async def process_audio(audio_file_content: bytes, lang1: str, lang2: str):
     """
@@ -117,10 +136,26 @@ async def process_audio(audio_file_content: bytes, lang1: str, lang2: str):
              return {"error": f"Detected language '{detected_lang_code}' was not one of the expected languages: {lang1} or {lang2}."}
 
         # --- Transcription ---
-        options = whisper.DecodingOptions(language=detected_lang_code, fp16=(device=="cuda"))
-        result = whisper.decode(model, mel, options)
-        transcribed_text = result.text
-        logger.info(f"Transcription: {transcribed_text}")
+        # Force Hindi transcription if detected language is Hindi
+        if detected_lang_code == "hi":
+            options = whisper.DecodingOptions(language="hi", fp16=(device=="cuda"))
+            result = whisper.decode(model, mel, options)
+            transcribed_text = result.text
+            logger.info(f"Transcription (forced Hindi): {transcribed_text}")
+            # If output is mostly Latin, retry with forced Hindi
+            latin_count = sum('a' <= c.lower() <= 'z' for c in transcribed_text)
+            devanagari_count = sum('\u0900' <= c <= '\u097F' for c in transcribed_text)
+            if latin_count > devanagari_count:
+                logger.info("Transcription appears to be in Latin script, retrying with forced Hindi language.")
+                options = whisper.DecodingOptions(language="hi", fp16=(device=="cuda"), task="transcribe")
+                result = whisper.decode(model, mel, options)
+                transcribed_text = result.text
+                logger.info(f"Retried Hindi transcription: {transcribed_text}")
+        else:
+            options = whisper.DecodingOptions(language=detected_lang_code, fp16=(device=="cuda"))
+            result = whisper.decode(model, mel, options)
+            transcribed_text = result.text
+            logger.info(f"Transcription: {transcribed_text}")
 
         # Determine the target language for translation
         target_lang = lang2 if detected_lang_code == lang1 else lang1