Spaces:
Build error
Build error
Update app.py
Browse files
app.py
CHANGED
|
@@ -9,6 +9,7 @@ import sys
|
|
| 9 |
from audio_processing import AudioProcessor
|
| 10 |
import spaces
|
| 11 |
from chunkedTranscriber import ChunkedTranscriber
|
|
|
|
| 12 |
|
| 13 |
|
| 14 |
logging.basicConfig(
|
|
@@ -18,6 +19,7 @@ logging.basicConfig(
|
|
| 18 |
)
|
| 19 |
logger = logging.getLogger(__name__)
|
| 20 |
|
|
|
|
| 21 |
def load_qa_model():
|
| 22 |
"""Load question-answering model with long context support."""
|
| 23 |
try:
|
|
@@ -153,279 +155,7 @@ def answer_question(context, question):
|
|
| 153 |
|
| 154 |
messages = [
|
| 155 |
# {"role": "system", "content": "You are a helpful assistant who can answer questions based on the given context."},
|
| 156 |
-
{"role":"system", "content":
|
| 157 |
-
Analyze a translated transcript of a conversation that may contain multiple speakers and summarize the information in a structured intelligence document.
|
| 158 |
-
|
| 159 |
-
The input format will include word-level or sentence-level timestamps, each indicating the speaker ID, language, and translated text.
|
| 160 |
-
|
| 161 |
-
# Input Format Overview
|
| 162 |
-
|
| 163 |
-
Word-Level Timestamps Example:
|
| 164 |
-
```
|
| 165 |
-
[Start Time - End Time] - Speaker <ID> - Language: <Translated Language> - Translated Text: "<Word>"
|
| 166 |
-
```
|
| 167 |
-
Example:
|
| 168 |
-
```
|
| 169 |
-
0.01-0.02 - Speaker 1 - Language: English - Translated Text: "Proceed"
|
| 170 |
-
0.02-0.025 - Speaker 1 - Language: English - Translated Text: "with"
|
| 171 |
-
0.025-0.032 - Speaker 2 - Language: English - Translated Text: "caution"
|
| 172 |
-
```
|
| 173 |
-
|
| 174 |
-
Optional Sentence-Level Structure Example:
|
| 175 |
-
```
|
| 176 |
-
[Start Time - End Time] - Speaker <ID> - Language: <Translated Language> - Translated Text: "<Sentence>"
|
| 177 |
-
```
|
| 178 |
-
Example with Sentence Grouping:
|
| 179 |
-
```
|
| 180 |
-
0.01-0.05 - Speaker 1 - Language: English - Translated Text: "Proceed with caution."
|
| 181 |
-
0.06-0.12 - Speaker 2 - Language: English - Translated Text: "All systems are ready."
|
| 182 |
-
```
|
| 183 |
-
|
| 184 |
-
# Intelligence Summary Document Structure
|
| 185 |
-
|
| 186 |
-
Use the format below to create a structured summary for each conversation transcript received:
|
| 187 |
-
|
| 188 |
-
### 1. Top-Level Status & Assessment:
|
| 189 |
-
- **Threat Level Assessment**:
|
| 190 |
-
- Choose one:
|
| 191 |
-
- Completely Innocuous
|
| 192 |
-
- Likely Innocuous
|
| 193 |
-
- Unclear — Requires Investigation
|
| 194 |
-
- Likely Dangerous — Immediate Action
|
| 195 |
-
- Likely Dangerous — Delayed Action
|
| 196 |
-
- 100% Dangerous — Immediate Action
|
| 197 |
-
- 100% Dangerous — Delayed Action
|
| 198 |
-
- **Humanitarian Alert**: Identify any indications of distress, coercion, or need for assistance, such as signs of duress or requests for help.
|
| 199 |
-
|
| 200 |
-
### 2. Basic Metadata:
|
| 201 |
-
- **Number of Speakers**: Total and unique speakers detected.
|
| 202 |
-
- **Languages**: List of languages used, with indication of who spoke which language.
|
| 203 |
-
- **Location**: Actual or inferred locations of participants.
|
| 204 |
-
- **Communication Medium**: Identify the method of interaction (e.g., phone call, direct conversation).
|
| 205 |
-
|
| 206 |
-
### 3. Conversation Overview:
|
| 207 |
-
- **Summary**: Concise breakdown of the main points and context.
|
| 208 |
-
- **Alarming Keywords**: Identify any concerning words, including but not limited to keywords like "kill," "attack," "weapon," etc.
|
| 209 |
-
- **Suspicious or Cryptic Phrases**: Statements that appear coded or unclear in the context of the discussion.
|
| 210 |
-
|
| 211 |
-
### 4. In-Depth Analysis:
|
| 212 |
-
- **Network Connections**: Identify mentions of additional individuals or groups involved.
|
| 213 |
-
- **Intent & Emotional Tone Detection**: Analyze emotional cues (e.g., anger, fear, calmness, urgency). Identify signs of deception or tension.
|
| 214 |
-
- **Behavioral Patterns**: Highlight repeated themes, phrases, or signals of planning and coordination.
|
| 215 |
-
- **Code Words & Cryptic Language**: Detect terms that may indicate hidden or covert meaning.
|
| 216 |
-
- **Geolocation References**: Point out any inferences regarding regional language or place names.
|
| 217 |
-
- **Sentiment on Strategic Issues**: Identify any indication of radical, dissenting, or anti-national views that could imply unrest or extremism.
|
| 218 |
-
|
| 219 |
-
### 5. Resource Mentions & Operational Logistics:
|
| 220 |
-
- **Resource & Asset Mentions**: List any mention of tools, weapons, vehicles, or supply logistics.
|
| 221 |
-
- **Behavioral Deviations**: Identify shifts in tone, speech, or demeanor suggesting stress, coercion, urgency, or preparation.
|
| 222 |
-
|
| 223 |
-
### 6. Prioritization, Recommendations & Actionables:
|
| 224 |
-
- **High-Risk Alert Priority**: Identify whether the conversation should be flagged for further attention.
|
| 225 |
-
- **Recommended Actions**:
|
| 226 |
-
- **Surveillance**: Suggest surveillance if concerning patterns or keywords are detected.
|
| 227 |
-
- **Intervention**: Recommend intervention for urgent/high-risk cases.
|
| 228 |
-
- **Humanitarian Assistance**: Suggest immediate support for any signs of distress.
|
| 229 |
-
- **Follow-Up Analysis**: Identify statements that need deeper review for clarity or to understand potential hidden meanings.
|
| 230 |
-
|
| 231 |
-
# Steps
|
| 232 |
-
|
| 233 |
-
1. Analyze the input conversation for participant information and context.
|
| 234 |
-
2. Fill in each section of the Intelligence Summary Document structure.
|
| 235 |
-
3. Ensure all details, especially those related to potential risk factors or alerts, are captured and highlighted clearly.
|
| 236 |
-
|
| 237 |
-
# Output Format
|
| 238 |
-
|
| 239 |
-
Provide one structured Intelligence Summary Document for the conversation in either plain text format or structured JSON.
|
| 240 |
-
|
| 241 |
-
# JSON Format Example:
|
| 242 |
-
```json
|
| 243 |
-
{
|
| 244 |
-
"Top-Level Status & Assessment": {
|
| 245 |
-
"Threat Level Assessment": "Unclear - Requires Investigation",
|
| 246 |
-
"Humanitarian Alert": "No distress signals detected."
|
| 247 |
-
},
|
| 248 |
-
"Basic Metadata": {
|
| 249 |
-
"Number of Speakers": 2,
|
| 250 |
-
"Languages": {
|
| 251 |
-
"Speaker 1": "English",
|
| 252 |
-
"Speaker 2": "English"
|
| 253 |
-
},
|
| 254 |
-
"Location": "Unknown",
|
| 255 |
-
"Communication Medium": "Direct conversation"
|
| 256 |
-
},
|
| 257 |
-
"Conversation Overview": {
|
| 258 |
-
"Summary": "A cautious approach was suggested by Speaker 1, followed by an assurance from Speaker 2 that systems are ready.",
|
| 259 |
-
"Alarming Keywords": [],
|
| 260 |
-
"Suspicious or Cryptic Phrases": []
|
| 261 |
-
},
|
| 262 |
-
"In-Depth Analysis": {
|
| 263 |
-
"Network Connections": "None identified",
|
| 264 |
-
"Intent & Emotional Tone Detection": "Calm, precautionary tone",
|
| 265 |
-
"Behavioral Patterns": "Speaker 1 expressing concern, Speaker 2 providing assurance",
|
| 266 |
-
"Code Words & Cryptic Language": [],
|
| 267 |
-
"Geolocation References": [],
|
| 268 |
-
"Sentiment on Strategic Issues": "No radical or dissenting sentiment detected"
|
| 269 |
-
},
|
| 270 |
-
"Resource Mentions & Operational Logistics": {
|
| 271 |
-
"Resource & Asset Mentions": [],
|
| 272 |
-
"Behavioral Deviations": "None noted"
|
| 273 |
-
},
|
| 274 |
-
"Prioritization, Recommendations & Actionables": {
|
| 275 |
-
"High-Risk Alert Priority": "Low",
|
| 276 |
-
"Recommended Actions": {
|
| 277 |
-
"Surveillance": "No further surveillance needed.",
|
| 278 |
-
"Intervention": "Not required.",
|
| 279 |
-
"Humanitarian Assistance": "Not required.",
|
| 280 |
-
"Follow-Up Analysis": "No unusual phrases detected requiring review."
|
| 281 |
-
}
|
| 282 |
-
}
|
| 283 |
-
}
|
| 284 |
-
```
|
| 285 |
-
|
| 286 |
-
# Notes
|
| 287 |
-
|
| 288 |
-
- Ensure that you mark any ambiguous segments as requiring further investigation.
|
| 289 |
-
- Pay attention to emotional tone shifts or sudden changes in behavior.
|
| 290 |
-
- If any direct or implied threat is detected, prioritize appropriately using the provided classifications.
|
| 291 |
-
- Err on the side of caution. In case there is even a remote possibility that there might be something that required human attention, flag it.
|
| 292 |
-
Analyze a translated transcript of a conversation that may contain multiple speakers and summarize the information in a structured intelligence document.
|
| 293 |
-
|
| 294 |
-
The input format will include word-level or sentence-level timestamps, each indicating the speaker ID, language, and translated text.
|
| 295 |
-
|
| 296 |
-
# Input Format Overview
|
| 297 |
-
|
| 298 |
-
Word-Level Timestamps Example:
|
| 299 |
-
```
|
| 300 |
-
[Start Time - End Time] - Speaker <ID> - Language: <Translated Language> - Translated Text: "<Word>"
|
| 301 |
-
```
|
| 302 |
-
Example:
|
| 303 |
-
```
|
| 304 |
-
0.01-0.02 - Speaker 1 - Language: English - Translated Text: "Proceed"
|
| 305 |
-
0.02-0.025 - Speaker 1 - Language: English - Translated Text: "with"
|
| 306 |
-
0.025-0.032 - Speaker 2 - Language: English - Translated Text: "caution"
|
| 307 |
-
```
|
| 308 |
-
|
| 309 |
-
Optional Sentence-Level Structure Example:
|
| 310 |
-
```
|
| 311 |
-
[Start Time - End Time] - Speaker <ID> - Language: <Translated Language> - Translated Text: "<Sentence>"
|
| 312 |
-
```
|
| 313 |
-
Example with Sentence Grouping:
|
| 314 |
-
```
|
| 315 |
-
0.01-0.05 - Speaker 1 - Language: English - Translated Text: "Proceed with caution."
|
| 316 |
-
0.06-0.12 - Speaker 2 - Language: English - Translated Text: "All systems are ready."
|
| 317 |
-
```
|
| 318 |
-
|
| 319 |
-
# Intelligence Summary Document Structure
|
| 320 |
-
|
| 321 |
-
Use the format below to create a structured summary for each conversation transcript received:
|
| 322 |
-
|
| 323 |
-
### 1. Top-Level Status & Assessment:
|
| 324 |
-
- **Threat Level Assessment**:
|
| 325 |
-
- Choose one:
|
| 326 |
-
- Completely Innocuous
|
| 327 |
-
- Likely Innocuous
|
| 328 |
-
- Unclear — Requires Investigation
|
| 329 |
-
- Likely Dangerous — Immediate Action
|
| 330 |
-
- Likely Dangerous — Delayed Action
|
| 331 |
-
- 100% Dangerous — Immediate Action
|
| 332 |
-
- 100% Dangerous — Delayed Action
|
| 333 |
-
- **Humanitarian Alert**: Identify any indications of distress, coercion, or need for assistance, such as signs of duress or requests for help.
|
| 334 |
-
|
| 335 |
-
### 2. Basic Metadata:
|
| 336 |
-
- **Number of Speakers**: Total and unique speakers detected.
|
| 337 |
-
- **Languages**: List of languages used, with indication of who spoke which language.
|
| 338 |
-
- **Location**: Actual or inferred locations of participants.
|
| 339 |
-
- **Communication Medium**: Identify the method of interaction (e.g., phone call, direct conversation).
|
| 340 |
-
|
| 341 |
-
### 3. Conversation Overview:
|
| 342 |
-
- **Summary**: Concise breakdown of the main points and context.
|
| 343 |
-
- **Alarming Keywords**: Identify any concerning words, including but not limited to keywords like "kill," "attack," "weapon," etc.
|
| 344 |
-
- **Suspicious or Cryptic Phrases**: Statements that appear coded or unclear in the context of the discussion.
|
| 345 |
-
|
| 346 |
-
### 4. In-Depth Analysis:
|
| 347 |
-
- **Network Connections**: Identify mentions of additional individuals or groups involved.
|
| 348 |
-
- **Intent & Emotional Tone Detection**: Analyze emotional cues (e.g., anger, fear, calmness, urgency). Identify signs of deception or tension.
|
| 349 |
-
- **Behavioral Patterns**: Highlight repeated themes, phrases, or signals of planning and coordination.
|
| 350 |
-
- **Code Words & Cryptic Language**: Detect terms that may indicate hidden or covert meaning.
|
| 351 |
-
- **Geolocation References**: Point out any inferences regarding regional language or place names.
|
| 352 |
-
- **Sentiment on Strategic Issues**: Identify any indication of radical, dissenting, or anti-national views that could imply unrest or extremism.
|
| 353 |
-
|
| 354 |
-
### 5. Resource Mentions & Operational Logistics:
|
| 355 |
-
- **Resource & Asset Mentions**: List any mention of tools, weapons, vehicles, or supply logistics.
|
| 356 |
-
- **Behavioral Deviations**: Identify shifts in tone, speech, or demeanor suggesting stress, coercion, urgency, or preparation.
|
| 357 |
-
|
| 358 |
-
### 6. Prioritization, Recommendations & Actionables:
|
| 359 |
-
- **High-Risk Alert Priority**: Identify whether the conversation should be flagged for further attention.
|
| 360 |
-
- **Recommended Actions**:
|
| 361 |
-
- **Surveillance**: Suggest surveillance if concerning patterns or keywords are detected.
|
| 362 |
-
- **Intervention**: Recommend intervention for urgent/high-risk cases.
|
| 363 |
-
- **Humanitarian Assistance**: Suggest immediate support for any signs of distress.
|
| 364 |
-
- **Follow-Up Analysis**: Identify statements that need deeper review for clarity or to understand potential hidden meanings.
|
| 365 |
-
|
| 366 |
-
# Steps
|
| 367 |
-
|
| 368 |
-
1. Analyze the input conversation for participant information and context.
|
| 369 |
-
2. Fill in each section of the Intelligence Summary Document structure.
|
| 370 |
-
3. Ensure all details, especially those related to potential risk factors or alerts, are captured and highlighted clearly.
|
| 371 |
-
|
| 372 |
-
# Output Format
|
| 373 |
-
|
| 374 |
-
Provide one structured Intelligence Summary Document for the conversation in either plain text format or structured JSON.
|
| 375 |
-
|
| 376 |
-
# JSON Format Example:
|
| 377 |
-
```json
|
| 378 |
-
{
|
| 379 |
-
"Top-Level Status & Assessment": {
|
| 380 |
-
"Threat Level Assessment": "Unclear - Requires Investigation",
|
| 381 |
-
"Humanitarian Alert": "No distress signals detected."
|
| 382 |
-
},
|
| 383 |
-
"Basic Metadata": {
|
| 384 |
-
"Number of Speakers": 2,
|
| 385 |
-
"Languages": {
|
| 386 |
-
"Speaker 1": "English",
|
| 387 |
-
"Speaker 2": "English"
|
| 388 |
-
},
|
| 389 |
-
"Location": "Unknown",
|
| 390 |
-
"Communication Medium": "Direct conversation"
|
| 391 |
-
},
|
| 392 |
-
"Conversation Overview": {
|
| 393 |
-
"Summary": "A cautious approach was suggested by Speaker 1, followed by an assurance from Speaker 2 that systems are ready.",
|
| 394 |
-
"Alarming Keywords": [],
|
| 395 |
-
"Suspicious or Cryptic Phrases": []
|
| 396 |
-
},
|
| 397 |
-
"In-Depth Analysis": {
|
| 398 |
-
"Network Connections": "None identified",
|
| 399 |
-
"Intent & Emotional Tone Detection": "Calm, precautionary tone",
|
| 400 |
-
"Behavioral Patterns": "Speaker 1 expressing concern, Speaker 2 providing assurance",
|
| 401 |
-
"Code Words & Cryptic Language": [],
|
| 402 |
-
"Geolocation References": [],
|
| 403 |
-
"Sentiment on Strategic Issues": "No radical or dissenting sentiment detected"
|
| 404 |
-
},
|
| 405 |
-
"Resource Mentions & Operational Logistics": {
|
| 406 |
-
"Resource & Asset Mentions": [],
|
| 407 |
-
"Behavioral Deviations": "None noted"
|
| 408 |
-
},
|
| 409 |
-
"Prioritization, Recommendations & Actionables": {
|
| 410 |
-
"High-Risk Alert Priority": "Low",
|
| 411 |
-
"Recommended Actions": {
|
| 412 |
-
"Surveillance": "No further surveillance needed.",
|
| 413 |
-
"Intervention": "Not required.",
|
| 414 |
-
"Humanitarian Assistance": "Not required.",
|
| 415 |
-
"Follow-Up Analysis": "No unusual phrases detected requiring review."
|
| 416 |
-
}
|
| 417 |
-
}
|
| 418 |
-
}
|
| 419 |
-
```
|
| 420 |
-
|
| 421 |
-
# Notes
|
| 422 |
-
|
| 423 |
-
- Ensure that you mark any ambiguous segments as requiring further investigation.
|
| 424 |
-
- Pay attention to emotional tone shifts or sudden changes in behavior.
|
| 425 |
-
- If any direct or implied threat is detected, prioritize appropriately using the provided classifications.
|
| 426 |
-
- Err on the side of caution. In case there is even a remote possibility that there might be something that required human attention, flag it.
|
| 427 |
-
|
| 428 |
-
"""},
|
| 429 |
{"role": "user", "content": f"Context: {text}\n\nQuestion: {question}"}
|
| 430 |
]
|
| 431 |
|
|
|
|
| 9 |
from audio_processing import AudioProcessor
|
| 10 |
import spaces
|
| 11 |
from chunkedTranscriber import ChunkedTranscriber
|
| 12 |
+
from system_message import SYSTEM_MESSAGE
|
| 13 |
|
| 14 |
|
| 15 |
logging.basicConfig(
|
|
|
|
| 19 |
)
|
| 20 |
logger = logging.getLogger(__name__)
|
| 21 |
|
| 22 |
+
|
| 23 |
def load_qa_model():
|
| 24 |
"""Load question-answering model with long context support."""
|
| 25 |
try:
|
|
|
|
| 155 |
|
| 156 |
messages = [
|
| 157 |
# {"role": "system", "content": "You are a helpful assistant who can answer questions based on the given context."},
|
| 158 |
+
{"role":"system", "content": SYSTEM_MESSAGE},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 159 |
{"role": "user", "content": f"Context: {text}\n\nQuestion: {question}"}
|
| 160 |
]
|
| 161 |
|