Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -9,13 +9,11 @@ from gtts import gTTS
|
|
| 9 |
import tempfile
|
| 10 |
import requests
|
| 11 |
|
| 12 |
-
|
| 13 |
# --- CONFIGURATION ---
|
| 14 |
MODEL_K2H_REPO = "ankitklakra/kurukh-to-hindi"
|
| 15 |
MODEL_H2K_REPO = "ankitklakra/hindi-to-kurukh"
|
| 16 |
SHEET_NAME = "Kurukh_Feedback_Log"
|
| 17 |
|
| 18 |
-
|
| 19 |
# --- LOAD MODELS ---
|
| 20 |
print(f"๐ Starting App with Gradio Version: {gr.__version__}")
|
| 21 |
|
|
@@ -30,20 +28,16 @@ pipe_h2k = pipeline("text2text-generation", model=model_h2k, tokenizer=tokenizer
|
|
| 30 |
print("Loading Voice Model...")
|
| 31 |
asr_pipeline = pipeline("automatic-speech-recognition", model="openai/whisper-tiny")
|
| 32 |
|
| 33 |
-
|
| 34 |
# --- HELPER FUNCTIONS ---
|
| 35 |
def transliterate_to_hindi(text):
|
| 36 |
try:
|
| 37 |
-
url =
|
| 38 |
-
"https://inputtools.google.com/request?text={}&itc=hi-t-i0-und&num=1"
|
| 39 |
-
)
|
| 40 |
response = requests.get(url.format(text))
|
| 41 |
result = response.json()
|
| 42 |
return result[1][0][1][0]
|
| 43 |
except:
|
| 44 |
return text
|
| 45 |
|
| 46 |
-
|
| 47 |
def save_to_sheet(original, translation, correction, direction):
|
| 48 |
try:
|
| 49 |
json_creds = os.getenv("GOOGLE_CREDENTIALS")
|
|
@@ -55,10 +49,9 @@ def save_to_sheet(original, translation, correction, direction):
|
|
| 55 |
"https://spreadsheets.google.com/feeds",
|
| 56 |
"https://www.googleapis.com/auth/drive",
|
| 57 |
]
|
| 58 |
-
creds = ServiceAccountCredentials.from_json_keyfile_dict(
|
| 59 |
-
creds_dict, scope
|
| 60 |
-
)
|
| 61 |
client = gspread.authorize(creds)
|
|
|
|
| 62 |
sheet = client.open(SHEET_NAME).sheet1
|
| 63 |
|
| 64 |
if not sheet.get_all_values():
|
|
@@ -73,56 +66,39 @@ def save_to_sheet(original, translation, correction, direction):
|
|
| 73 |
)
|
| 74 |
|
| 75 |
sheet.append_row(
|
| 76 |
-
[
|
| 77 |
-
str(datetime.now()),
|
| 78 |
-
direction,
|
| 79 |
-
original,
|
| 80 |
-
translation,
|
| 81 |
-
correction,
|
| 82 |
-
]
|
| 83 |
)
|
|
|
|
| 84 |
return "โ
Saved to Google Sheets."
|
| 85 |
except Exception as e:
|
| 86 |
return f"โ Error: {str(e)}"
|
| 87 |
|
| 88 |
-
|
| 89 |
def speech_to_text(audio_path):
|
| 90 |
if audio_path is None:
|
| 91 |
return ""
|
| 92 |
return asr_pipeline(audio_path)["text"]
|
| 93 |
|
| 94 |
-
|
| 95 |
def text_to_speech(text, language="hi"):
|
| 96 |
if not text:
|
| 97 |
return None
|
| 98 |
try:
|
| 99 |
tts = gTTS(text=text, lang=language)
|
| 100 |
-
temp_file = tempfile.NamedTemporaryFile(
|
| 101 |
-
delete=False, suffix=".mp3"
|
| 102 |
-
)
|
| 103 |
tts.save(temp_file.name)
|
| 104 |
return temp_file.name
|
| 105 |
except:
|
| 106 |
return None
|
| 107 |
|
| 108 |
-
|
| 109 |
# --- MAIN TRANSLATION LOGIC ---
|
| 110 |
def process_translation(text, audio_input, direction, is_hinglish):
|
| 111 |
-
original_text = (
|
| 112 |
-
speech_to_text(audio_input) if audio_input else text
|
| 113 |
-
)
|
| 114 |
-
|
| 115 |
if not original_text:
|
| 116 |
return "", "", None
|
| 117 |
|
| 118 |
if direction == "Hindi -> Kurukh" and is_hinglish:
|
| 119 |
original_text = transliterate_to_hindi(original_text)
|
| 120 |
|
| 121 |
-
target_pipeline =
|
| 122 |
-
pipe_k2h
|
| 123 |
-
if direction == "Kurukh -> Hindi"
|
| 124 |
-
else pipe_h2k
|
| 125 |
-
)
|
| 126 |
|
| 127 |
try:
|
| 128 |
results = target_pipeline(
|
|
@@ -144,71 +120,62 @@ def process_translation(text, audio_input, direction, is_hinglish):
|
|
| 144 |
return original_text, translated_text, audio_output
|
| 145 |
|
| 146 |
|
| 147 |
-
# ---
|
| 148 |
universal_css = """
|
| 149 |
<style>
|
| 150 |
@import url('https://fonts.googleapis.com/css2?family=Poppins:wght@300;400;600&display=swap');
|
| 151 |
-
|
| 152 |
-
|
| 153 |
-
font-family: 'Poppins', sans-serif !important;
|
| 154 |
}
|
| 155 |
-
|
| 156 |
-
|
| 157 |
-
|
| 158 |
-
margin-bottom: 25px;
|
| 159 |
padding: 20px;
|
| 160 |
background: linear-gradient(to right, #f8f9fa, #e9ecef);
|
| 161 |
border-radius: 15px;
|
| 162 |
}
|
| 163 |
-
.header-title {
|
| 164 |
-
font-size: 2.2em;
|
| 165 |
-
font-weight: 700;
|
| 166 |
-
color: #2c3e50;
|
| 167 |
}
|
| 168 |
-
.header-subtitle {
|
| 169 |
-
font-size: 1.1em;
|
| 170 |
-
color: #576574;
|
| 171 |
}
|
| 172 |
</style>
|
| 173 |
"""
|
| 174 |
|
| 175 |
|
| 176 |
# --- UI ---
|
| 177 |
-
with gr.Blocks(
|
| 178 |
-
title="Kurukh AI Translator", theme=gr.themes.Soft()
|
| 179 |
-
) as demo:
|
| 180 |
|
| 181 |
gr.HTML(universal_css)
|
| 182 |
|
| 183 |
-
gr.HTML(
|
| 184 |
-
"""
|
| 185 |
<div class="header-div">
|
| 186 |
<h1 class="header-title">๐ฎ๐ณ AI Kurukh (Oraon) Translator</h1>
|
| 187 |
<p class="header-subtitle">
|
| 188 |
Bridging Communities with Artificial Intelligence | Voice & Hinglish Supported
|
| 189 |
</p>
|
| 190 |
</div>
|
| 191 |
-
"""
|
| 192 |
-
)
|
| 193 |
|
| 194 |
with gr.Tabs():
|
| 195 |
|
| 196 |
-
#
|
| 197 |
-
with gr.
|
| 198 |
|
| 199 |
-
with gr.Accordion(
|
| 200 |
-
|
| 201 |
-
|
| 202 |
-
|
| 203 |
-
|
| 204 |
-
|
| 205 |
-
2. **Hinglish:** Enable if you type Hindi in English
|
| 206 |
-
3. **Voice:** Speak using the microphone (Hindi only)
|
| 207 |
-
"""
|
| 208 |
-
)
|
| 209 |
|
| 210 |
with gr.Row():
|
| 211 |
|
|
|
|
| 212 |
with gr.Column():
|
| 213 |
direction = gr.Radio(
|
| 214 |
["Kurukh -> Hindi", "Hindi -> Kurukh"],
|
|
@@ -217,55 +184,42 @@ with gr.Blocks(
|
|
| 217 |
)
|
| 218 |
|
| 219 |
is_hinglish = gr.Checkbox(
|
| 220 |
-
label="๐ค Hinglish Typing", value=False
|
| 221 |
)
|
| 222 |
|
| 223 |
input_text = gr.Textbox(
|
| 224 |
-
label="Enter Text",
|
| 225 |
-
placeholder="Type sentences here...",
|
| 226 |
-
lines=4,
|
| 227 |
)
|
| 228 |
|
| 229 |
input_audio = gr.Audio(
|
| 230 |
-
label="๐๏ธ Voice Input (Hindi Only)",
|
| 231 |
sources=["microphone"],
|
| 232 |
type="filepath",
|
|
|
|
| 233 |
)
|
| 234 |
|
| 235 |
-
translate_btn = gr.Button(
|
| 236 |
-
"Translate ๐", variant="primary"
|
| 237 |
-
)
|
| 238 |
|
|
|
|
| 239 |
with gr.Column():
|
| 240 |
output_text = gr.Textbox(
|
| 241 |
label="Translation",
|
| 242 |
lines=4,
|
| 243 |
interactive=False,
|
|
|
|
| 244 |
)
|
| 245 |
-
|
| 246 |
output_audio = gr.Audio(
|
| 247 |
-
label="๐ Listen (Hindi Only)",
|
| 248 |
-
interactive=False,
|
| 249 |
-
autoplay=False,
|
| 250 |
)
|
| 251 |
|
| 252 |
translate_btn.click(
|
| 253 |
fn=process_translation,
|
| 254 |
-
inputs=[
|
| 255 |
-
input_text,
|
| 256 |
-
input_audio,
|
| 257 |
-
direction,
|
| 258 |
-
is_hinglish,
|
| 259 |
-
],
|
| 260 |
outputs=[input_text, output_text, output_audio],
|
| 261 |
)
|
| 262 |
|
| 263 |
-
#
|
| 264 |
-
with gr.
|
| 265 |
-
|
| 266 |
-
gr.Markdown(
|
| 267 |
-
"### ๐ ๏ธ Help us improve accuracy by submitting corrections."
|
| 268 |
-
)
|
| 269 |
|
| 270 |
fb_direction = gr.Radio(
|
| 271 |
["Kurukh -> Hindi", "Hindi -> Kurukh"],
|
|
@@ -274,34 +228,23 @@ with gr.Blocks(
|
|
| 274 |
)
|
| 275 |
|
| 276 |
fb_original = gr.Textbox(label="Original Text")
|
| 277 |
-
fb_ai_output = gr.Textbox(
|
| 278 |
-
label="AI's Wrong Translation"
|
| 279 |
-
)
|
| 280 |
fb_user_correct = gr.Textbox(
|
| 281 |
-
label="
|
| 282 |
-
lines=2,
|
| 283 |
-
)
|
| 284 |
-
|
| 285 |
-
submit_btn = gr.Button(
|
| 286 |
-
"Submit Correction to Database", variant="secondary"
|
| 287 |
)
|
| 288 |
|
|
|
|
| 289 |
status_lbl = gr.Label(label="Status")
|
| 290 |
|
| 291 |
submit_btn.click(
|
| 292 |
fn=save_to_sheet,
|
| 293 |
-
inputs=[
|
| 294 |
-
fb_original,
|
| 295 |
-
fb_ai_output,
|
| 296 |
-
fb_user_correct,
|
| 297 |
-
fb_direction,
|
| 298 |
-
],
|
| 299 |
outputs=status_lbl,
|
| 300 |
)
|
| 301 |
|
| 302 |
gr.Markdown("---")
|
| 303 |
gr.HTML(
|
| 304 |
-
"<center style='color:#
|
| 305 |
)
|
| 306 |
|
| 307 |
demo.launch()
|
|
|
|
| 9 |
import tempfile
|
| 10 |
import requests
|
| 11 |
|
|
|
|
| 12 |
# --- CONFIGURATION ---
|
| 13 |
MODEL_K2H_REPO = "ankitklakra/kurukh-to-hindi"
|
| 14 |
MODEL_H2K_REPO = "ankitklakra/hindi-to-kurukh"
|
| 15 |
SHEET_NAME = "Kurukh_Feedback_Log"
|
| 16 |
|
|
|
|
| 17 |
# --- LOAD MODELS ---
|
| 18 |
print(f"๐ Starting App with Gradio Version: {gr.__version__}")
|
| 19 |
|
|
|
|
| 28 |
print("Loading Voice Model...")
|
| 29 |
asr_pipeline = pipeline("automatic-speech-recognition", model="openai/whisper-tiny")
|
| 30 |
|
|
|
|
| 31 |
# --- HELPER FUNCTIONS ---
|
| 32 |
def transliterate_to_hindi(text):
|
| 33 |
try:
|
| 34 |
+
url = "https://inputtools.google.com/request?text={}&itc=hi-t-i0-und&num=1"
|
|
|
|
|
|
|
| 35 |
response = requests.get(url.format(text))
|
| 36 |
result = response.json()
|
| 37 |
return result[1][0][1][0]
|
| 38 |
except:
|
| 39 |
return text
|
| 40 |
|
|
|
|
| 41 |
def save_to_sheet(original, translation, correction, direction):
|
| 42 |
try:
|
| 43 |
json_creds = os.getenv("GOOGLE_CREDENTIALS")
|
|
|
|
| 49 |
"https://spreadsheets.google.com/feeds",
|
| 50 |
"https://www.googleapis.com/auth/drive",
|
| 51 |
]
|
| 52 |
+
creds = ServiceAccountCredentials.from_json_keyfile_dict(creds_dict, scope)
|
|
|
|
|
|
|
| 53 |
client = gspread.authorize(creds)
|
| 54 |
+
|
| 55 |
sheet = client.open(SHEET_NAME).sheet1
|
| 56 |
|
| 57 |
if not sheet.get_all_values():
|
|
|
|
| 66 |
)
|
| 67 |
|
| 68 |
sheet.append_row(
|
| 69 |
+
[str(datetime.now()), direction, original, translation, correction]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 70 |
)
|
| 71 |
+
|
| 72 |
return "โ
Saved to Google Sheets."
|
| 73 |
except Exception as e:
|
| 74 |
return f"โ Error: {str(e)}"
|
| 75 |
|
|
|
|
| 76 |
def speech_to_text(audio_path):
|
| 77 |
if audio_path is None:
|
| 78 |
return ""
|
| 79 |
return asr_pipeline(audio_path)["text"]
|
| 80 |
|
|
|
|
| 81 |
def text_to_speech(text, language="hi"):
|
| 82 |
if not text:
|
| 83 |
return None
|
| 84 |
try:
|
| 85 |
tts = gTTS(text=text, lang=language)
|
| 86 |
+
temp_file = tempfile.NamedTemporaryFile(delete=False, suffix=".mp3")
|
|
|
|
|
|
|
| 87 |
tts.save(temp_file.name)
|
| 88 |
return temp_file.name
|
| 89 |
except:
|
| 90 |
return None
|
| 91 |
|
|
|
|
| 92 |
# --- MAIN TRANSLATION LOGIC ---
|
| 93 |
def process_translation(text, audio_input, direction, is_hinglish):
|
| 94 |
+
original_text = speech_to_text(audio_input) if audio_input else text
|
|
|
|
|
|
|
|
|
|
| 95 |
if not original_text:
|
| 96 |
return "", "", None
|
| 97 |
|
| 98 |
if direction == "Hindi -> Kurukh" and is_hinglish:
|
| 99 |
original_text = transliterate_to_hindi(original_text)
|
| 100 |
|
| 101 |
+
target_pipeline = pipe_k2h if direction == "Kurukh -> Hindi" else pipe_h2k
|
|
|
|
|
|
|
|
|
|
|
|
|
| 102 |
|
| 103 |
try:
|
| 104 |
results = target_pipeline(
|
|
|
|
| 120 |
return original_text, translated_text, audio_output
|
| 121 |
|
| 122 |
|
| 123 |
+
# --- CSS ---
|
| 124 |
universal_css = """
|
| 125 |
<style>
|
| 126 |
@import url('https://fonts.googleapis.com/css2?family=Poppins:wght@300;400;600&display=swap');
|
| 127 |
+
body, button, input, select, textarea, .gradio-container {
|
| 128 |
+
font-family: 'Poppins', sans-serif !important;
|
|
|
|
| 129 |
}
|
| 130 |
+
.header-div {
|
| 131 |
+
text-align: center;
|
| 132 |
+
margin-bottom: 25px;
|
|
|
|
| 133 |
padding: 20px;
|
| 134 |
background: linear-gradient(to right, #f8f9fa, #e9ecef);
|
| 135 |
border-radius: 15px;
|
| 136 |
}
|
| 137 |
+
.header-title {
|
| 138 |
+
font-size: 2.2em;
|
| 139 |
+
font-weight: 700;
|
| 140 |
+
color: #2c3e50;
|
| 141 |
}
|
| 142 |
+
.header-subtitle {
|
| 143 |
+
font-size: 1.1em;
|
| 144 |
+
color: #576574;
|
| 145 |
}
|
| 146 |
</style>
|
| 147 |
"""
|
| 148 |
|
| 149 |
|
| 150 |
# --- UI ---
|
| 151 |
+
with gr.Blocks(title="Kurukh AI Translator") as demo:
|
|
|
|
|
|
|
| 152 |
|
| 153 |
gr.HTML(universal_css)
|
| 154 |
|
| 155 |
+
gr.HTML("""
|
|
|
|
| 156 |
<div class="header-div">
|
| 157 |
<h1 class="header-title">๐ฎ๐ณ AI Kurukh (Oraon) Translator</h1>
|
| 158 |
<p class="header-subtitle">
|
| 159 |
Bridging Communities with Artificial Intelligence | Voice & Hinglish Supported
|
| 160 |
</p>
|
| 161 |
</div>
|
| 162 |
+
""")
|
|
|
|
| 163 |
|
| 164 |
with gr.Tabs():
|
| 165 |
|
| 166 |
+
# --- Translator Tab ---
|
| 167 |
+
with gr.TabItem("๐ฃ๏ธ Translator"):
|
| 168 |
|
| 169 |
+
with gr.Accordion("โน๏ธ How to use (Click to expand)", open=False):
|
| 170 |
+
gr.Markdown("""
|
| 171 |
+
1. Select translation mode.
|
| 172 |
+
2. Enable Hinglish if typing Hindi in English letters.
|
| 173 |
+
3. Use Voice input if needed.
|
| 174 |
+
""")
|
|
|
|
|
|
|
|
|
|
|
|
|
| 175 |
|
| 176 |
with gr.Row():
|
| 177 |
|
| 178 |
+
# LEFT
|
| 179 |
with gr.Column():
|
| 180 |
direction = gr.Radio(
|
| 181 |
["Kurukh -> Hindi", "Hindi -> Kurukh"],
|
|
|
|
| 184 |
)
|
| 185 |
|
| 186 |
is_hinglish = gr.Checkbox(
|
| 187 |
+
label="๐ค Hinglish Typing (e.g., 'Tumhara')", value=False
|
| 188 |
)
|
| 189 |
|
| 190 |
input_text = gr.Textbox(
|
| 191 |
+
label="Enter Text", placeholder="Type sentences here...", lines=4
|
|
|
|
|
|
|
| 192 |
)
|
| 193 |
|
| 194 |
input_audio = gr.Audio(
|
|
|
|
| 195 |
sources=["microphone"],
|
| 196 |
type="filepath",
|
| 197 |
+
label="๐๏ธ Voice Input (Hindi Only)",
|
| 198 |
)
|
| 199 |
|
| 200 |
+
translate_btn = gr.Button("Translate ๐")
|
|
|
|
|
|
|
| 201 |
|
| 202 |
+
# RIGHT
|
| 203 |
with gr.Column():
|
| 204 |
output_text = gr.Textbox(
|
| 205 |
label="Translation",
|
| 206 |
lines=4,
|
| 207 |
interactive=False,
|
| 208 |
+
show_copy_button=True,
|
| 209 |
)
|
|
|
|
| 210 |
output_audio = gr.Audio(
|
| 211 |
+
label="๐ Listen (Hindi Only)", interactive=False
|
|
|
|
|
|
|
| 212 |
)
|
| 213 |
|
| 214 |
translate_btn.click(
|
| 215 |
fn=process_translation,
|
| 216 |
+
inputs=[input_text, input_audio, direction, is_hinglish],
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 217 |
outputs=[input_text, output_text, output_audio],
|
| 218 |
)
|
| 219 |
|
| 220 |
+
# --- Feedback Tab ---
|
| 221 |
+
with gr.TabItem("๐ Improve the AI"):
|
| 222 |
+
gr.Markdown("### ๐ ๏ธ Help us improve accuracy")
|
|
|
|
|
|
|
|
|
|
| 223 |
|
| 224 |
fb_direction = gr.Radio(
|
| 225 |
["Kurukh -> Hindi", "Hindi -> Kurukh"],
|
|
|
|
| 228 |
)
|
| 229 |
|
| 230 |
fb_original = gr.Textbox(label="Original Text")
|
| 231 |
+
fb_ai_output = gr.Textbox(label="AI's Translation")
|
|
|
|
|
|
|
| 232 |
fb_user_correct = gr.Textbox(
|
| 233 |
+
label="Correct Translation", lines=2
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 234 |
)
|
| 235 |
|
| 236 |
+
submit_btn = gr.Button("Submit Correction")
|
| 237 |
status_lbl = gr.Label(label="Status")
|
| 238 |
|
| 239 |
submit_btn.click(
|
| 240 |
fn=save_to_sheet,
|
| 241 |
+
inputs=[fb_original, fb_ai_output, fb_user_correct, fb_direction],
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 242 |
outputs=status_lbl,
|
| 243 |
)
|
| 244 |
|
| 245 |
gr.Markdown("---")
|
| 246 |
gr.HTML(
|
| 247 |
+
"<center style='color:#777;'>Built with โค๏ธ for the Kurukh Community</center>"
|
| 248 |
)
|
| 249 |
|
| 250 |
demo.launch()
|