Spaces:
Sleeping
Sleeping
- .env +9 -0
- app.py +235 -70
- llm_client.py +77 -0
- prompts.py +65 -0
- readme.md +0 -0
- requirements.txt +4 -0
.env
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
SPEECH_KEY="4CcN3sox1gfqI81AOhFHIYZwGusC6frPa1kSO32gIjjPSCFxke0EJQQJ99CBACYeBjFXJ3w3AAAYACOGd9oM"
|
| 2 |
+
SPEECH_ENDPOINT="https://eastus.api.cognitive.microsoft.com/"
|
| 3 |
+
SPEECH_REGION="eastus"
|
| 4 |
+
|
| 5 |
+
# Azure OpenAI
|
| 6 |
+
AZURE_OPENAI_KEY="8wKFXqTCFBDBZ8eMj1ePBxVF0XMUbH9H50XXuV3ReJ0ZpAMrRfCcJQQJ99BEACHYHv6XJ3w3AAAAACOGqdEB"
|
| 7 |
+
AZURE_OPENAI_ENDPOINT="https://teera-maz475y3-eastus2.cognitiveservices.azure.com/"
|
| 8 |
+
AZURE_OPENAI_API_VERSION="2024-12-01-preview"
|
| 9 |
+
AZURE_OPENAI_DEPLOYMENT="gpt-5.2-chat"
|
app.py
CHANGED
|
@@ -1,70 +1,235 @@
|
|
| 1 |
-
import
|
| 2 |
-
|
| 3 |
-
|
| 4 |
-
|
| 5 |
-
|
| 6 |
-
|
| 7 |
-
|
| 8 |
-
|
| 9 |
-
|
| 10 |
-
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
|
| 14 |
-
|
| 15 |
-
|
| 16 |
-
|
| 17 |
-
|
| 18 |
-
|
| 19 |
-
|
| 20 |
-
|
| 21 |
-
|
| 22 |
-
|
| 23 |
-
|
| 24 |
-
|
| 25 |
-
|
| 26 |
-
|
| 27 |
-
|
| 28 |
-
|
| 29 |
-
|
| 30 |
-
|
| 31 |
-
|
| 32 |
-
|
| 33 |
-
|
| 34 |
-
|
| 35 |
-
|
| 36 |
-
|
| 37 |
-
|
| 38 |
-
|
| 39 |
-
|
| 40 |
-
|
| 41 |
-
|
| 42 |
-
|
| 43 |
-
""
|
| 44 |
-
|
| 45 |
-
|
| 46 |
-
|
| 47 |
-
|
| 48 |
-
|
| 49 |
-
|
| 50 |
-
|
| 51 |
-
|
| 52 |
-
|
| 53 |
-
|
| 54 |
-
|
| 55 |
-
|
| 56 |
-
|
| 57 |
-
|
| 58 |
-
|
| 59 |
-
|
| 60 |
-
|
| 61 |
-
|
| 62 |
-
|
| 63 |
-
|
| 64 |
-
|
| 65 |
-
|
| 66 |
-
|
| 67 |
-
|
| 68 |
-
|
| 69 |
-
|
| 70 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
import sys
|
| 3 |
+
import azure.cognitiveservices.speech as speechsdk
|
| 4 |
+
from dotenv import load_dotenv
|
| 5 |
+
|
| 6 |
+
load_dotenv()
|
| 7 |
+
|
| 8 |
+
SPEECH_KEY = os.getenv("SPEECH_KEY")
|
| 9 |
+
SPEECH_REGION = os.getenv("SPEECH_REGION", "eastus")
|
| 10 |
+
|
| 11 |
+
|
| 12 |
+
def create_speech_config(language="th-TH"):
|
| 13 |
+
"""Create a SpeechConfig with the given language."""
|
| 14 |
+
config = speechsdk.SpeechConfig(
|
| 15 |
+
subscription=SPEECH_KEY,
|
| 16 |
+
region=SPEECH_REGION,
|
| 17 |
+
)
|
| 18 |
+
config.speech_recognition_language = language
|
| 19 |
+
return config
|
| 20 |
+
|
| 21 |
+
|
| 22 |
+
def transcribe_from_mic():
|
| 23 |
+
"""Transcribe from the local microphone (CLI mode)."""
|
| 24 |
+
speech_config = create_speech_config("th-TH")
|
| 25 |
+
audio_config = speechsdk.audio.AudioConfig(use_default_microphone=True)
|
| 26 |
+
recognizer = speechsdk.SpeechRecognizer(
|
| 27 |
+
speech_config=speech_config,
|
| 28 |
+
audio_config=audio_config,
|
| 29 |
+
)
|
| 30 |
+
|
| 31 |
+
print("🎤 Listening... Speak into your microphone.")
|
| 32 |
+
result = recognizer.recognize_once()
|
| 33 |
+
|
| 34 |
+
if result.reason == speechsdk.ResultReason.RecognizedSpeech:
|
| 35 |
+
print("✅ Recognized: " + result.text)
|
| 36 |
+
elif result.reason == speechsdk.ResultReason.NoMatch:
|
| 37 |
+
print("❌ No speech could be recognized: " + str(result.no_match_details))
|
| 38 |
+
elif result.reason == speechsdk.ResultReason.Canceled:
|
| 39 |
+
cancellation_details = result.cancellation_details
|
| 40 |
+
print("⚠️ Speech recognition canceled: " + str(cancellation_details.reason))
|
| 41 |
+
if cancellation_details.reason == speechsdk.CancellationReason.Error:
|
| 42 |
+
print("Error details: " + str(cancellation_details.error_details))
|
| 43 |
+
print("Did you set the speech resource key and region?")
|
| 44 |
+
|
| 45 |
+
|
| 46 |
+
def transcribe_audio_file(audio_path, language="th-TH"):
|
| 47 |
+
"""Transcribe an audio file using Azure Speech SDK."""
|
| 48 |
+
if audio_path is None:
|
| 49 |
+
return "⚠️ กรุณาอัดเสียงก่อน"
|
| 50 |
+
|
| 51 |
+
speech_config = create_speech_config(language)
|
| 52 |
+
audio_config = speechsdk.audio.AudioConfig(filename=audio_path)
|
| 53 |
+
recognizer = speechsdk.SpeechRecognizer(
|
| 54 |
+
speech_config=speech_config,
|
| 55 |
+
audio_config=audio_config,
|
| 56 |
+
)
|
| 57 |
+
|
| 58 |
+
# Use continuous recognition to get the full transcript
|
| 59 |
+
all_results = []
|
| 60 |
+
done = False
|
| 61 |
+
|
| 62 |
+
def on_recognized(evt):
|
| 63 |
+
if evt.result.reason == speechsdk.ResultReason.RecognizedSpeech:
|
| 64 |
+
all_results.append(evt.result.text)
|
| 65 |
+
|
| 66 |
+
def on_canceled(evt):
|
| 67 |
+
nonlocal done
|
| 68 |
+
done = True
|
| 69 |
+
|
| 70 |
+
def on_stopped(evt):
|
| 71 |
+
nonlocal done
|
| 72 |
+
done = True
|
| 73 |
+
|
| 74 |
+
recognizer.recognized.connect(on_recognized)
|
| 75 |
+
recognizer.canceled.connect(on_canceled)
|
| 76 |
+
recognizer.session_stopped.connect(on_stopped)
|
| 77 |
+
|
| 78 |
+
recognizer.start_continuous_recognition()
|
| 79 |
+
|
| 80 |
+
import time
|
| 81 |
+
while not done:
|
| 82 |
+
time.sleep(0.1)
|
| 83 |
+
|
| 84 |
+
recognizer.stop_continuous_recognition()
|
| 85 |
+
|
| 86 |
+
if all_results:
|
| 87 |
+
return "\n".join(all_results)
|
| 88 |
+
else:
|
| 89 |
+
return "❌ ไม่สามารถถอดเสียงได้ — ลองพูดดังขึ้นหรือตรวจสอบไมค์"
|
| 90 |
+
|
| 91 |
+
|
| 92 |
+
def transcribe_and_analyze(audio_path, language):
|
| 93 |
+
"""Transcribe audio, then analyze with LLM. Returns (transcript, analysis_json)."""
|
| 94 |
+
transcript = transcribe_audio_file(audio_path, language)
|
| 95 |
+
|
| 96 |
+
if transcript.startswith("❌") or transcript.startswith("⚠️"):
|
| 97 |
+
return transcript, ""
|
| 98 |
+
|
| 99 |
+
from llm_client import analyze_football_content, format_analysis_result
|
| 100 |
+
result = analyze_football_content(transcript)
|
| 101 |
+
analysis_json = format_analysis_result(result)
|
| 102 |
+
|
| 103 |
+
return transcript, analysis_json
|
| 104 |
+
|
| 105 |
+
|
| 106 |
+
def analyze_text_only(transcript):
|
| 107 |
+
"""Analyze existing transcript text without re-transcribing."""
|
| 108 |
+
if not transcript or not transcript.strip():
|
| 109 |
+
return "⚠️ กรุณาใส่ข้อความก่อน"
|
| 110 |
+
|
| 111 |
+
from llm_client import analyze_football_content, format_analysis_result
|
| 112 |
+
result = analyze_football_content(transcript)
|
| 113 |
+
return format_analysis_result(result)
|
| 114 |
+
|
| 115 |
+
|
| 116 |
+
def run_web():
|
| 117 |
+
"""Run the Gradio web UI."""
|
| 118 |
+
import gradio as gr
|
| 119 |
+
|
| 120 |
+
with gr.Blocks(
|
| 121 |
+
title="ASR - Football Analysis",
|
| 122 |
+
theme=gr.themes.Soft(
|
| 123 |
+
primary_hue=gr.themes.colors.indigo,
|
| 124 |
+
secondary_hue=gr.themes.colors.purple,
|
| 125 |
+
neutral_hue=gr.themes.colors.slate,
|
| 126 |
+
),
|
| 127 |
+
css="""
|
| 128 |
+
.gradio-container {
|
| 129 |
+
max-width: 900px !important;
|
| 130 |
+
margin: auto !important;
|
| 131 |
+
}
|
| 132 |
+
""",
|
| 133 |
+
) as app:
|
| 134 |
+
|
| 135 |
+
gr.Markdown(
|
| 136 |
+
"""
|
| 137 |
+
# ⚽ Football Speech Analyzer
|
| 138 |
+
### ถอดเสียงพูด + วิเคราะห์เนื้อหาฟุตบอลด้วย AI
|
| 139 |
+
---
|
| 140 |
+
"""
|
| 141 |
+
)
|
| 142 |
+
|
| 143 |
+
with gr.Row():
|
| 144 |
+
language = gr.Dropdown(
|
| 145 |
+
choices=[
|
| 146 |
+
("🇹🇭 ไทย", "th-TH"),
|
| 147 |
+
("🇺🇸 English", "en-US"),
|
| 148 |
+
("🇯🇵 日本語", "ja-JP"),
|
| 149 |
+
("🇨🇳 中文", "zh-CN"),
|
| 150 |
+
("🇰🇷 한국어", "ko-KR"),
|
| 151 |
+
],
|
| 152 |
+
value="th-TH",
|
| 153 |
+
label="ภาษา",
|
| 154 |
+
interactive=True,
|
| 155 |
+
)
|
| 156 |
+
|
| 157 |
+
gr.Markdown("### 🎤 อัดเสียงจากไมค์")
|
| 158 |
+
audio_input = gr.Audio(
|
| 159 |
+
sources=["microphone", "upload"],
|
| 160 |
+
type="filepath",
|
| 161 |
+
label="กดปุ่มอัดเสียง หรืออัปโหลดไฟล์เสียง",
|
| 162 |
+
)
|
| 163 |
+
|
| 164 |
+
with gr.Row():
|
| 165 |
+
transcribe_btn = gr.Button(
|
| 166 |
+
"✨ ถอดเสียงอย่างเดียว",
|
| 167 |
+
variant="secondary",
|
| 168 |
+
size="lg",
|
| 169 |
+
)
|
| 170 |
+
full_btn = gr.Button(
|
| 171 |
+
"⚽ ถอดเสียง + วิเคราะห์ฟุตบอล",
|
| 172 |
+
variant="primary",
|
| 173 |
+
size="lg",
|
| 174 |
+
)
|
| 175 |
+
|
| 176 |
+
gr.Markdown("### 📝 ข้อความที่ถอดได้")
|
| 177 |
+
output_text = gr.Textbox(
|
| 178 |
+
label="Transcript",
|
| 179 |
+
lines=6,
|
| 180 |
+
show_copy_button=True,
|
| 181 |
+
placeholder="ผลการถอดเสียงจะแสดงที่นี่...",
|
| 182 |
+
)
|
| 183 |
+
|
| 184 |
+
gr.Markdown("### 🧠 ผลวิเคราะห์จาก AI")
|
| 185 |
+
with gr.Row():
|
| 186 |
+
analyze_btn = gr.Button(
|
| 187 |
+
"🔄 วิเคราะห์ข้อความข้างบนอีกครั้ง",
|
| 188 |
+
variant="secondary",
|
| 189 |
+
size="sm",
|
| 190 |
+
)
|
| 191 |
+
|
| 192 |
+
analysis_output = gr.Code(
|
| 193 |
+
label="Football Analysis (JSON)",
|
| 194 |
+
language="json",
|
| 195 |
+
lines=20,
|
| 196 |
+
)
|
| 197 |
+
|
| 198 |
+
# --- Events ---
|
| 199 |
+
|
| 200 |
+
# Transcribe only
|
| 201 |
+
transcribe_btn.click(
|
| 202 |
+
fn=transcribe_audio_file,
|
| 203 |
+
inputs=[audio_input, language],
|
| 204 |
+
outputs=output_text,
|
| 205 |
+
)
|
| 206 |
+
|
| 207 |
+
# Transcribe + Analyze
|
| 208 |
+
full_btn.click(
|
| 209 |
+
fn=transcribe_and_analyze,
|
| 210 |
+
inputs=[audio_input, language],
|
| 211 |
+
outputs=[output_text, analysis_output],
|
| 212 |
+
)
|
| 213 |
+
|
| 214 |
+
# Re-analyze existing transcript
|
| 215 |
+
analyze_btn.click(
|
| 216 |
+
fn=analyze_text_only,
|
| 217 |
+
inputs=output_text,
|
| 218 |
+
outputs=analysis_output,
|
| 219 |
+
)
|
| 220 |
+
|
| 221 |
+
# Auto-transcribe + analyze on recording stop
|
| 222 |
+
audio_input.stop_recording(
|
| 223 |
+
fn=transcribe_and_analyze,
|
| 224 |
+
inputs=[audio_input, language],
|
| 225 |
+
outputs=[output_text, analysis_output],
|
| 226 |
+
)
|
| 227 |
+
|
| 228 |
+
app.launch(server_name="127.0.0.1", server_port=7860)
|
| 229 |
+
|
| 230 |
+
|
| 231 |
+
if __name__ == "__main__":
|
| 232 |
+
if "--cli" in sys.argv:
|
| 233 |
+
transcribe_from_mic()
|
| 234 |
+
else:
|
| 235 |
+
run_web()
|
llm_client.py
ADDED
|
@@ -0,0 +1,77 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
"""
|
| 2 |
+
Azure OpenAI client wrapper for LLM-based analysis.
|
| 3 |
+
"""
|
| 4 |
+
|
| 5 |
+
import os
|
| 6 |
+
import json
|
| 7 |
+
from openai import AzureOpenAI
|
| 8 |
+
from dotenv import load_dotenv
|
| 9 |
+
from prompts import FOOTBALL_ANALYSIS_SYSTEM_PROMPT, FOOTBALL_ANALYSIS_USER_PROMPT
|
| 10 |
+
|
| 11 |
+
load_dotenv()
|
| 12 |
+
|
| 13 |
+
AZURE_OPENAI_KEY = os.getenv("AZURE_OPENAI_KEY")
|
| 14 |
+
AZURE_OPENAI_ENDPOINT = os.getenv("AZURE_OPENAI_ENDPOINT")
|
| 15 |
+
AZURE_OPENAI_API_VERSION = os.getenv("AZURE_OPENAI_API_VERSION", "2024-12-01-preview")
|
| 16 |
+
AZURE_OPENAI_DEPLOYMENT = os.getenv("AZURE_OPENAI_DEPLOYMENT", "gpt-5.2-chat")
|
| 17 |
+
|
| 18 |
+
|
| 19 |
+
def get_openai_client():
|
| 20 |
+
"""Create and return an Azure OpenAI client."""
|
| 21 |
+
return AzureOpenAI(
|
| 22 |
+
api_key=AZURE_OPENAI_KEY,
|
| 23 |
+
azure_endpoint=AZURE_OPENAI_ENDPOINT,
|
| 24 |
+
api_version=AZURE_OPENAI_API_VERSION,
|
| 25 |
+
)
|
| 26 |
+
|
| 27 |
+
|
| 28 |
+
def analyze_football_content(transcript: str) -> dict:
|
| 29 |
+
"""
|
| 30 |
+
Send transcribed text to Azure OpenAI for football content analysis.
|
| 31 |
+
|
| 32 |
+
Args:
|
| 33 |
+
transcript: The transcribed speech text.
|
| 34 |
+
|
| 35 |
+
Returns:
|
| 36 |
+
A dict with categorized football data (teams, leagues, sentiment, etc.)
|
| 37 |
+
"""
|
| 38 |
+
if not transcript or not transcript.strip():
|
| 39 |
+
return {"error": "ไม่มีข้อความให้วิเคราะห์"}
|
| 40 |
+
|
| 41 |
+
client = get_openai_client()
|
| 42 |
+
|
| 43 |
+
user_message = FOOTBALL_ANALYSIS_USER_PROMPT.format(transcript=transcript)
|
| 44 |
+
|
| 45 |
+
try:
|
| 46 |
+
response = client.chat.completions.create(
|
| 47 |
+
model=AZURE_OPENAI_DEPLOYMENT,
|
| 48 |
+
messages=[
|
| 49 |
+
{"role": "system", "content": FOOTBALL_ANALYSIS_SYSTEM_PROMPT},
|
| 50 |
+
{"role": "user", "content": user_message},
|
| 51 |
+
],
|
| 52 |
+
max_completion_tokens=4096,
|
| 53 |
+
)
|
| 54 |
+
|
| 55 |
+
content = response.choices[0].message.content.strip()
|
| 56 |
+
|
| 57 |
+
# Clean up markdown code fences if the model wraps JSON in ```json ... ```
|
| 58 |
+
if content.startswith("```"):
|
| 59 |
+
content = content.split("\n", 1)[1] # Remove first line (```json)
|
| 60 |
+
content = content.rsplit("```", 1)[0] # Remove last ```
|
| 61 |
+
content = content.strip()
|
| 62 |
+
|
| 63 |
+
result = json.loads(content)
|
| 64 |
+
return result
|
| 65 |
+
|
| 66 |
+
except json.JSONDecodeError:
|
| 67 |
+
return {
|
| 68 |
+
"error": "LLM ตอบกลับมาไม่ใช่ JSON ที่ถูกต้อง",
|
| 69 |
+
"raw_response": content,
|
| 70 |
+
}
|
| 71 |
+
except Exception as e:
|
| 72 |
+
return {"error": f"เกิดข้อผิดพลาด: {str(e)}"}
|
| 73 |
+
|
| 74 |
+
|
| 75 |
+
def format_analysis_result(result: dict) -> str:
|
| 76 |
+
"""Format the analysis result as a pretty JSON string for display."""
|
| 77 |
+
return json.dumps(result, ensure_ascii=False, indent=2)
|
prompts.py
ADDED
|
@@ -0,0 +1,65 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
"""
|
| 2 |
+
Prompt templates for LLM-based football content analysis.
|
| 3 |
+
"""
|
| 4 |
+
|
| 5 |
+
FOOTBALL_ANALYSIS_SYSTEM_PROMPT = """คุณเป็น AI ผู้เชี่ยวชาญด้านการวิเคราะห์เนื้อหาฟุตบอล
|
| 6 |
+
หน้าที่ของคุณคือวิเคราะห์ข้อความที่ถอดเสียงมา (transcript) แล้วจัดหมวดหมู่ข้อมูลเกี่ยวกับฟุตบอล
|
| 7 |
+
|
| 8 |
+
คุณต้องตอบกลับเป็น JSON เท่านั้น ตามโครงสร้างนี้:
|
| 9 |
+
|
| 10 |
+
{
|
| 11 |
+
"teams_mentioned": [
|
| 12 |
+
{
|
| 13 |
+
"name": "ชื่อทีม (ภาษาอังกฤษ)",
|
| 14 |
+
"name_th": "ชื่อทีม (ภาษาไทย ถ้ามี)",
|
| 15 |
+
"context": "บริบทที่พูดถึงทีมนี้โดยย่อ"
|
| 16 |
+
}
|
| 17 |
+
],
|
| 18 |
+
"leagues_mentioned": [
|
| 19 |
+
{
|
| 20 |
+
"name": "ชื่อลีก (ภาษาอังกฤษ)",
|
| 21 |
+
"name_th": "ชื่อลีก (ภาษาไทย ถ้ามี)",
|
| 22 |
+
"country": "ประเทศ"
|
| 23 |
+
}
|
| 24 |
+
],
|
| 25 |
+
"players_mentioned": [
|
| 26 |
+
{
|
| 27 |
+
"name": "ชื่อนักเตะ",
|
| 28 |
+
"team": "ทีมที่สังกัด (ถ้าระบุได้)",
|
| 29 |
+
"context": "บริบทที่พูดถึง"
|
| 30 |
+
}
|
| 31 |
+
],
|
| 32 |
+
"topics": ["หัวข้อที่พูดถึง เช่น ผลการแข่งขัน, ตลาดการย้ายทีม, อาการบาดเจ็บ, ..."],
|
| 33 |
+
"sentiment": {
|
| 34 |
+
"overall": "positive | negative | neutral | mixed",
|
| 35 |
+
"score": 0.0,
|
| 36 |
+
"details": "อธิบายเหตุผลโดยย่อ"
|
| 37 |
+
},
|
| 38 |
+
"match_info": {
|
| 39 |
+
"is_match_discussed": true,
|
| 40 |
+
"home_team": "ทีมเหย้า (ถ้ามี)",
|
| 41 |
+
"away_team": "ทีมเยือน (ถ้ามี)",
|
| 42 |
+
"score": "ผลสกอร์ (ถ้ามี)",
|
| 43 |
+
"competition": "รายการแข่งขัน (ถ้ามี)"
|
| 44 |
+
},
|
| 45 |
+
"summary": "สรุปเนื้อหาโดยย่อ 1-2 ประโยค",
|
| 46 |
+
"confidence": 0.0,
|
| 47 |
+
"is_football_content": true
|
| 48 |
+
}
|
| 49 |
+
|
| 50 |
+
กฎ:
|
| 51 |
+
1. ถ้าเนื้อหาไม่เกี่ยวกับฟุตบอลเลย ให้ตั้ง "is_football_content" เป็น false และใส่ข้อมูลที่เกี่ยวข้องน้อยที่สุด
|
| 52 |
+
2. "sentiment.score" อยู่ในช่วง -1.0 (ลบมาก) ถึง 1.0 (บวกมาก), 0.0 คือ neutral
|
| 53 |
+
3. "confidence" อยู่ในช่วง 0.0-1.0 แสดงความมั่นใจในการวิเคราะห์
|
| 54 |
+
4. ตอบเป็น JSON เท่านั้น ห้ามมีข้อความอื่นนอกเหนือจาก JSON
|
| 55 |
+
5. ถ้าไม่มีข้อมูลในฟิลด์ใด ให้ใส่ null หรือ array ว่าง []
|
| 56 |
+
6. พยายามระบุชื่อเป็นภาษาอังกฤษมาตรฐานเสมอ (เช่น "Liverpool", ไม่ใช่ "ลิเวอร์พูล" อย่างเดียว)
|
| 57 |
+
"""
|
| 58 |
+
|
| 59 |
+
FOOTBALL_ANALYSIS_USER_PROMPT = """วิเคราะห์ข้อความที่ถอดเสียงมานี้:
|
| 60 |
+
|
| 61 |
+
---
|
| 62 |
+
{transcript}
|
| 63 |
+
---
|
| 64 |
+
|
| 65 |
+
ตอบเป็น JSON ตามโครงสร้างที่กำหนด"""
|
readme.md
ADDED
|
File without changes
|
requirements.txt
ADDED
|
@@ -0,0 +1,4 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
azure-cognitiveservices-speech
|
| 2 |
+
python-dotenv
|
| 3 |
+
gradio
|
| 4 |
+
openai
|