Spaces:
Sleeping
Sleeping
Upload 3 files
Browse files- app.py +108 -0
- config.json +1 -0
- requirements.txt +3 -0
app.py
ADDED
|
@@ -0,0 +1,108 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
#1. Import libraries
|
| 2 |
+
import requests
|
| 3 |
+
import json
|
| 4 |
+
import yt_dlp
|
| 5 |
+
import time # Delay ditambahkan agar tidak terlalu cepat request ke API
|
| 6 |
+
|
| 7 |
+
#2. Load API KEY
|
| 8 |
+
def load_api_key():
|
| 9 |
+
try:
|
| 10 |
+
with open("config.json", "r") as f:
|
| 11 |
+
config = json.load(f)
|
| 12 |
+
return config.get("API_KEY", None)
|
| 13 |
+
except FileNotFoundError:
|
| 14 |
+
print("Error: config.json tidak ditemukan! Harap tambahkan API_KEY.")
|
| 15 |
+
exit()
|
| 16 |
+
except json.JSONDecodeError:
|
| 17 |
+
print("Error: config.json tidak valid!")
|
| 18 |
+
exit()
|
| 19 |
+
|
| 20 |
+
API_KEY = load_api_key()
|
| 21 |
+
if not API_KEY:
|
| 22 |
+
print("Error: API_KEY tidak ditemukan dalam config.json")
|
| 23 |
+
exit()
|
| 24 |
+
|
| 25 |
+
YOUTUBE_URL = "https://www.youtube.com/watch?v=_xIwjmCH6D4"
|
| 26 |
+
|
| 27 |
+
#3. Ambil audio dari URL YouTube
|
| 28 |
+
def get_audio_url(youtube_url):
|
| 29 |
+
ydl_opts = {
|
| 30 |
+
"format": "bestaudio/best", # Download hanya audio
|
| 31 |
+
"postprocessors": [{"key": "FFmpegExtractAudio", "preferredcodec": "mp3"}], # Convert ke MP3
|
| 32 |
+
"noplaylist": True, # Hanya download satu video
|
| 33 |
+
"quiet": True, # Hilangkan log yang tidak perlu
|
| 34 |
+
"threads": 4 # Gunakan multi-threading
|
| 35 |
+
}
|
| 36 |
+
|
| 37 |
+
with yt_dlp.YoutubeDL(ydl_opts) as ydl:
|
| 38 |
+
try:
|
| 39 |
+
info_dict = ydl.extract_info(youtube_url, download=False)
|
| 40 |
+
return info_dict.get('url', None)
|
| 41 |
+
except Exception as e:
|
| 42 |
+
print(f"Error: Gagal mengambil audio dari YouTube. {e}")
|
| 43 |
+
return None
|
| 44 |
+
|
| 45 |
+
#4. Upload audio ke AssemblyAI
|
| 46 |
+
def upload_audio(audio_url):
|
| 47 |
+
if not audio_url:
|
| 48 |
+
print("Error: Tidak ada audio yang bisa diupload.")
|
| 49 |
+
return None
|
| 50 |
+
|
| 51 |
+
headers = {"authorization": API_KEY, "content-type": "application/json"}
|
| 52 |
+
try:
|
| 53 |
+
response = requests.post(
|
| 54 |
+
"https://api.assemblyai.com/v2/transcript",
|
| 55 |
+
json={"audio_url": audio_url},
|
| 56 |
+
headers=headers
|
| 57 |
+
)
|
| 58 |
+
response_json = response.json()
|
| 59 |
+
return response_json.get('id', None)
|
| 60 |
+
except requests.RequestException as e:
|
| 61 |
+
print(f"Error: Gagal upload audio ke AssemblyAI. {e}")
|
| 62 |
+
return None
|
| 63 |
+
|
| 64 |
+
#5. Ambil hasil transkripsi
|
| 65 |
+
def get_transcription(transcript_id):
|
| 66 |
+
if not transcript_id:
|
| 67 |
+
return "Error: Tidak ada ID transkripsi."
|
| 68 |
+
|
| 69 |
+
headers = {'authorization': API_KEY}
|
| 70 |
+
url = f"https://api.assemblyai.com/v2/transcript/{transcript_id}"
|
| 71 |
+
|
| 72 |
+
while True:
|
| 73 |
+
try:
|
| 74 |
+
response = requests.get(url, headers=headers)
|
| 75 |
+
response_json = response.json()
|
| 76 |
+
status = response_json.get('status', None)
|
| 77 |
+
|
| 78 |
+
if status == 'completed':
|
| 79 |
+
return response_json.get('text', "Transkripsi tidak ditemukan.")
|
| 80 |
+
elif status == 'failed':
|
| 81 |
+
return "Transcription failed!"
|
| 82 |
+
else:
|
| 83 |
+
print("Menunggu transkripsi selesai...")
|
| 84 |
+
time.sleep(5) # Tambahkan delay agar tidak spam request
|
| 85 |
+
|
| 86 |
+
except requests.RequestException as e:
|
| 87 |
+
print(f"Error: Gagal mengambil transkripsi. {e}")
|
| 88 |
+
return None
|
| 89 |
+
|
| 90 |
+
#6. Jalankan semua proses
|
| 91 |
+
print("\n=== Memproses Video ===")
|
| 92 |
+
audio_url = get_audio_url(YOUTUBE_URL)
|
| 93 |
+
if not audio_url:
|
| 94 |
+
print("Gagal mengambil audio dari video.")
|
| 95 |
+
exit()
|
| 96 |
+
|
| 97 |
+
print("Mengupload audio ke AssemblyAI...")
|
| 98 |
+
transcript_id = upload_audio(audio_url)
|
| 99 |
+
if not transcript_id:
|
| 100 |
+
print("Gagal upload audio ke AssemblyAI.")
|
| 101 |
+
exit()
|
| 102 |
+
|
| 103 |
+
print("Menunggu hasil transkripsi...")
|
| 104 |
+
transcription = get_transcription(transcript_id)
|
| 105 |
+
|
| 106 |
+
#7. Print hasil transkripsi
|
| 107 |
+
print("\n=== Transkripsi Video ===\n")
|
| 108 |
+
print(transcription)
|
config.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"API_KEY": "3a5c38e449bf4d238390a5c3e1bbb5ae"}
|
requirements.txt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
streamlit
|
| 2 |
+
yt-dlp
|
| 3 |
+
assemblyai
|