Upload 3 files

Hinglish_Profanity_List.csv

@@ -0,0 +1,211 @@
+profanity_hn,profanity_en,ind
+badir,idiot,1
+badirchand,idiot,1
+bakland,idiot,1
+bhadva,pimp,2
+bhootnika,son of a witch,3
+chinaal,whore,3
+chup,shut up,1
+chutia,fucker ,5
+ghasti,hooker,4
+chutiya,fucker,5
+haraami,bastard,5
+haraam,bastard,5
+hijra,transsexual ,3
+hinjda,transsexual ,3
+jaanvar,animal,1
+kutta,dog,2
+kutiya,bitch,3
+khota,donkey,1
+auladheen,sonless,2
+jaat,breed,1
+najayaz,illegitimate,3
+gandpaidaish,badborn,2
+saala,sister�s husband,2
+kutti,bitch,2
+soover,swine,3
+tatti,shit,3
+potty,shit,3
+bahenchod,sister fucker,9
+bahanchod,sister fucker,9
+bahencho,sister fucker,9
+bancho,sister fucker,7
+bhenchod,profanity,7
+bahenke,sister�s ,6
+laude,dick,7
+takke,balls,7
+betichod,daughter fucker,9
+bhaichod,brother fucker,8
+bhains,buffalo,1
+jhalla,faggot,6
+jhant,pubic,7
+nabaal,hairless,4
+pissu,bug,3
+kutte,dog,2
+maadherchod,mother fucker,10
+madarchod,motherfucker,10
+padma,fat bitch,6
+raand,whore,9
+jamai,son-in-law,5
+randwa,male prostitute,8
+randi,hooker,8
+bachachod,son fucker,8
+bachichod,daughter fucker,8
+soower,swine,3
+bachchechod,children fucker,8
+ullu,idiot,1
+pathe,idiot,1
+banda,semi-dick,7
+booblay,boobs,7
+booby,boobs,7
+buble,boobs,7
+babla,boobs,7
+bhonsriwala,fucker,8
+bhonsdiwala,fucker,8
+ched,pussy,8
+chut,pussy,8
+chod,fuck,7
+chodu,fucker,7
+chodra,fucker,7
+choochi,boobs,8
+chuchi,boobs,8
+gaandu,asshole,9
+gandu,asshole,9
+gaand,ass,9
+lavda, dick ,8
+lawda,dick,8
+lauda,dick,8
+lund�, dick,8
+balchod,hair fucker,7
+lavander,dick head,8
+muth,masturbate ,7
+maacho,mother fucker,10
+mammey,boobs,8
+tatte,boobs,8
+toto,penis,8
+toota,broken,7
+backar,gossip,2
+bhandwe,pimp,7
+bhosadchod,ass fucker,9
+bhosad,pussy,8
+bumchod,ass fucker,9
+bum,ass,8
+bur,pussy,9
+chatani,ketchup,1
+cunt,pussy,8
+cuntmama,pussy,8
+chipkali,lizzard,1
+pasine,sweat,1
+jhaat,cunt,8
+chodela,fucked up,6
+bhagatchod,saint fucker,7
+chhola,clit,7
+chudai,fucking,7
+chudaikhana,whore house,9
+chunni,clit,8
+choot,pussy,8
+bhoot,ghost,1
+dhakkan,idiot,1
+bhajiye,snack,1
+fateychu,torn pussy,9
+gandnatije,Bad result,2
+lundtopi,condom,4
+gaandu,ass,8
+gaandfat,ass,8
+gaandmasti,ass,8
+makhanchudai,fucking,8
+gaandmarau,ass fuck,9
+gandu,faggot,8
+chaatu,licker,6
+beej,semen,6
+choosu,sucker,7
+fakeerchod,saint fucker,8
+lundoos,dick,8
+shorba,semen,7
+binbheja,brainless,3
+bhadwe,pimp,6
+parichod,angel fucker,9
+nirodh,condom.,5
+pucchi,pussy,8
+baajer,fucker,8
+choud,fuck,8
+bhosda,pussy,9
+sadi,stinking,5
+choos,suck,5
+maka,mother�s,7
+chinaal,prostitute,7
+gadde,boobs,7
+joon,bug,3
+chullugand,handful dirt,4
+doob,drown,1
+khatmal,bug,1
+gandkate,ass,6
+bambu,bamboo,4
+lassan,garlic,1
+danda,stick,2
+keera,bug,2
+keeda,bug,2
+hazaarchu,thousand pussy,7
+paidaishikeeda,born bug,5
+kali,nigger,5
+safaid,american,2
+poot,son,2
+behendi,sister,5
+chus,sucker,6
+machudi,mother fucker,10
+chodoonga,fuck,8
+baapchu,father pussy,9
+laltern,lantern,5
+suhaagchudai,wedding fuck,8
+raatchuda,night fuck,7
+kaalu,migga,6
+neech,low caste,7
+chikna,gay,6
+meetha,gay,6
+beechka,gay,6
+chooche,boobs,8
+patichod,husband,8
+rundi,prostitute,7
+makkhi,fly,1
+biwichod,wife fucker,9
+chodhunga,fuck,8
+haathi,elephant,1
+kute,dog,2
+jhanten,pubic hair,8
+kaat,cut,3
+gandi,filthy,3
+gadha,donkey,1
+bimaar,ill,2
+badboodar,smelly,2
+dum,tail,2
+raandsaala,sister�s brother pimp,7
+phudi,pussy,7
+chute,pussy,7
+kussi,ass,7
+khandanchod,family fucker,9
+ghussa,fuck,6
+maarey,dead,4
+chipkili,lizard,1
+unday,eggs,1
+budh,cunt,7
+chaarpai,cot,1
+chodun,fuck,5
+chatri,condom,3
+chode,fuck,6
+chodho,fuck,6
+mullekatue,Derogatory abuse to muslims,6
+mullikatui,Derogatory Abuse to female muslim,6
+mullekebaal,Derogatory Abuse to muslim,6
+momedankatue,Derogatory Abuse to muslim,6
+katua,dick cut,8
+chutiyapa,fuck all,8
+bc,sister fucker,10
+mc,mother fucker,10
+chudwaya,fuck,7
+kutton,dog,2
+jungli,wild,2
+vahiyaat,disgusting,4
+jihadi,terrorist,4
+atankvadi,terrorist,4
+atankwadi,terrorist,4
+aatanki,terorist,4

app.py

@@ -0,0 +1,102 @@
+from gradio_client import Client, handle_file
+import pandas as pd
+import gradio as gr
+from vosk import Model, KaldiRecognizer
+import json
+import wave
+
+clientEngText = Client("dj-dawgs-ipd/IPD-Text-English-Finetune")
+clientHingText = Client("dj-dawgs-ipd/IPD-Text-Hinglish")
+clientAud = Client("dj-dawgs-ipd/IPD_Audio_HuBERT")
+
+profanity_df = pd.read_csv('Hinglish_Profanity_List.csv', encoding='utf-8')
+profanity_hn = profanity_df['profanity_hn']
+vosk_model = Model(lang="en-us")
+
+
+# import whisper
+# def stt_whisper(file_path):
+#     model = whisper.load_model("base")
+#     try:
+#         result = model.transcribe(file_path)
+#         return result["text"]
+#     except Exception as e:
+#           print(e)
+#         return ""
+
+
+def stt_vosk(file_path):
+    try:
+        wf = wave.open(file_path, "rb")
+        rec = KaldiRecognizer(vosk_model, wf.getframerate())
+        rec.SetWords(True)
+        rec.SetPartialWords(True)
+        while True:
+            data = wf.readframes(4000)
+            if len(data) == 0:
+                break
+            rec.AcceptWaveform(data)
+        data = json.loads(rec.FinalResult())
+        return data["text"]
+    except:
+        return ""
+
+
+def extract_text(audio_path):
+    return stt_vosk(audio_path).lower()
+
+
+def predict_hate_speech(audio_path):
+
+    audResult = clientAud.predict(
+        audio_path=handle_file(audio_path),
+        api_name="/predict"
+    )
+    audResult = json.loads(audResult.replace("'", '"'))
+
+    stt_text = extract_text(audio_path)
+
+    engResult = clientEngText.predict(
+        text=stt_text[:200],
+        api_name="/predict"
+    )
+
+    hingResult = clientHingText.predict(
+        text=stt_text[:200],
+        api_name="/predict"
+    )
+
+    profanityFound = any(word in stt_text.split() for word in profanity_hn)
+    threshold = 0.6
+    isHate = (engResult[0] != "NEITHER" and engResult[1] > threshold) or (
+        hingResult[0] != "NAG" and hingResult[1] > threshold) or (
+        audResult['Classification'] == 'Hate Speech\n' and audResult['Confidence'] > threshold)
+
+    engConf = engResult[1] if engResult[0] != "NEITHER" else (1 - engResult[1])
+    hingConf = hingResult[1] if hingResult[0] != "NEITHER" else (1 - hingResult[1])
+    audConf = audResult['Confidence'] if audResult['Classification'] == 'Hate Speech\n' else (1 - audResult['Confidence'])
+
+    confidence = (engConf + hingConf + audConf) / 3
+    # print(profanityFound, engResult, hingResult, audResult)
+    if profanityFound:
+        return ["hate", f"Result: Profanity Found", f"Text: {stt_text}"]
+    elif isHate:
+        return ["hate", f"Confidence: {confidence}", f"Text: {stt_text}"]
+
+    return ["not_hate", "No hate found, yay!"]
+
+
+iface = gr.Interface(
+    fn=predict_hate_speech,
+    inputs=gr.Audio(type="filepath", label="Upload Audio"),
+    outputs=gr.Textbox(label="Hate Speech Analysis"),
+    title="Hate Speech Audio Pipeline",
+    description="Upload an audio file to detect potential hate speech content.",
+    examples=[
+        ["hate_video_3_3_snippet2.wav"]
+    ],
+    allow_flagging="manual"
+)
+
+if __name__ == "__main__":
+    iface.launch()

requirements.txt

@@ -0,0 +1,6 @@
+gradio
+gradio_client
+pandas
+vosk
+json
+wave