Spaces:
Build error
Build error
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,28 +1,9 @@
|
|
| 1 |
-
import
|
| 2 |
-
|
| 3 |
-
# Force upgrade huggingface_hub and transformers
|
| 4 |
-
subprocess.run(["pip", "install", "--upgrade", "huggingface_hub>=0.23.0", "transformers>=4.40.0"])
|
| 5 |
-
import subprocess
|
| 6 |
-
|
| 7 |
-
# Force upgrade huggingface_hub
|
| 8 |
-
subprocess.run(["pip", "install", "--upgrade", "huggingface_hub"])
|
| 9 |
-
import subprocess
|
| 10 |
-
subprocess.run(["pip", "install", "git+https://github.com/myshell-ai/OpenVoice.git"])
|
| 11 |
-
import os
|
| 12 |
-
from huggingface_hub import login
|
| 13 |
-
|
| 14 |
-
hf_token = os.getenv("HF_TOKEN")
|
| 15 |
-
|
| 16 |
-
if hf_token:
|
| 17 |
-
login(token=hf_token)
|
| 18 |
-
else:
|
| 19 |
-
print("β οΈ No HF_TOKEN found β some models may not load")
|
| 20 |
from pydub import AudioSegment
|
| 21 |
import numpy as np
|
| 22 |
import tempfile
|
| 23 |
import os
|
| 24 |
import noisereduce as nr
|
| 25 |
-
import json
|
| 26 |
import torch
|
| 27 |
from demucs import pretrained
|
| 28 |
from demucs.apply import apply_model
|
|
@@ -31,18 +12,15 @@ from pathlib import Path
|
|
| 31 |
import matplotlib.pyplot as plt
|
| 32 |
from io import BytesIO
|
| 33 |
from PIL import Image
|
| 34 |
-
import
|
|
|
|
|
|
|
| 35 |
import datetime
|
| 36 |
import librosa
|
| 37 |
import joblib
|
| 38 |
import warnings
|
| 39 |
-
from faster_whisper import WhisperModel
|
| 40 |
from mutagen.mp3 import MP3
|
| 41 |
from mutagen.id3 import ID3, TIT2, TPE1, TALB, TYER
|
| 42 |
-
import whisper
|
| 43 |
-
from pyannote.audio import Pipeline as DiarizationPipeline
|
| 44 |
-
from openvoice.api import TTS, ToneColorConverter
|
| 45 |
-
from openvoice.se_extractor import get_se
|
| 46 |
|
| 47 |
# Suppress warnings
|
| 48 |
warnings.filterwarnings("ignore")
|
|
@@ -147,7 +125,7 @@ def stem_split(audio_path):
|
|
| 147 |
|
| 148 |
return [gr.File(value=path) for path in stem_paths]
|
| 149 |
|
| 150 |
-
# ===
|
| 151 |
def load_presets():
|
| 152 |
try:
|
| 153 |
preset_files = [f for f in os.listdir("presets") if f.endswith(".json")]
|
|
@@ -177,7 +155,7 @@ if not preset_choices:
|
|
| 177 |
|
| 178 |
preset_names = list(preset_choices.keys())
|
| 179 |
|
| 180 |
-
# === Waveform
|
| 181 |
def show_waveform(audio_file):
|
| 182 |
try:
|
| 183 |
audio = AudioSegment.from_file(audio_file)
|
|
@@ -193,27 +171,18 @@ def show_waveform(audio_file):
|
|
| 193 |
except Exception as e:
|
| 194 |
return None
|
| 195 |
|
| 196 |
-
def detect_genre(audio_path):
|
| 197 |
-
try:
|
| 198 |
-
y, sr = torchaudio.load(audio_path)
|
| 199 |
-
mfccs = librosa.feature.mfcc(y=y.numpy().flatten(), sr=sr, n_mfcc=13).mean(axis=1).reshape(1, -1)
|
| 200 |
-
return "Speech"
|
| 201 |
-
except Exception:
|
| 202 |
-
return "Unknown"
|
| 203 |
-
|
| 204 |
# === Session Info Export ===
|
| 205 |
-
def generate_session_log(audio_path, effects, isolate_vocals, export_format
|
| 206 |
log = {
|
| 207 |
"timestamp": str(datetime.datetime.now()),
|
| 208 |
"filename": os.path.basename(audio_path),
|
| 209 |
"effects_applied": effects,
|
| 210 |
"isolate_vocals": isolate_vocals,
|
| 211 |
-
"export_format": export_format
|
| 212 |
-
"detected_genre": genre
|
| 213 |
}
|
| 214 |
return json.dumps(log, indent=2)
|
| 215 |
|
| 216 |
-
# === Main Processing Function
|
| 217 |
def process_audio(audio_file, selected_effects, isolate_vocals, preset_name, export_format):
|
| 218 |
status = "π Loading audio..."
|
| 219 |
try:
|
|
@@ -251,92 +220,73 @@ def process_audio(audio_file, selected_effects, isolate_vocals, preset_name, exp
|
|
| 251 |
final_audio.export(output_path, format=export_format.lower())
|
| 252 |
|
| 253 |
waveform_image = show_waveform(output_path)
|
| 254 |
-
|
| 255 |
-
session_log = generate_session_log(audio_file, effects_to_apply, isolate_vocals, export_format, genre)
|
| 256 |
|
| 257 |
status = "π Done!"
|
| 258 |
-
return output_path, waveform_image, session_log,
|
| 259 |
|
| 260 |
except Exception as e:
|
| 261 |
status = f"β Error: {str(e)}"
|
| 262 |
-
return None, None, status,
|
| 263 |
-
|
| 264 |
-
# === Batch Processing Function ===
|
| 265 |
-
def batch_process_audio(files, selected_effects, isolate_vocals, preset_name, export_format):
|
| 266 |
-
status = "π Loading files..."
|
| 267 |
-
try:
|
| 268 |
-
output_dir = tempfile.mkdtemp()
|
| 269 |
-
results = []
|
| 270 |
-
session_logs = []
|
| 271 |
-
|
| 272 |
-
for file in files:
|
| 273 |
-
processed_path, _, log, _, _ = process_audio(file.name, selected_effects, isolate_vocals, preset_name, export_format)
|
| 274 |
-
results.append(processed_path)
|
| 275 |
-
session_logs.append(log)
|
| 276 |
-
|
| 277 |
-
zip_path = os.path.join(output_dir, "batch_output.zip")
|
| 278 |
-
with zipfile.ZipFile(zip_path, 'w') as zipf:
|
| 279 |
-
for i, res in enumerate(results):
|
| 280 |
-
filename = f"processed_{i}.{export_format.lower()}"
|
| 281 |
-
zipf.write(res, filename)
|
| 282 |
-
zipf.writestr(f"session_info_{i}.json", session_logs[i])
|
| 283 |
-
|
| 284 |
-
return zip_path, "π¦ ZIP created successfully!"
|
| 285 |
|
| 286 |
-
|
| 287 |
-
return None, f"β Batch processing failed: {str(e)}"
|
| 288 |
-
|
| 289 |
-
# === Load Models Once at Start ===
|
| 290 |
-
|
| 291 |
-
# π§ Speaker Diarization Model
|
| 292 |
-
diarize_model = DiarizationPipeline.from_pretrained("pyannote/speaker-diarization", use_auth_token="YOUR_HF_TOKEN")
|
| 293 |
-
|
| 294 |
-
# π€ OpenVoice TTS + Converter
|
| 295 |
-
tts_model = TTS(lang='en')
|
| 296 |
-
tone_converter = ToneColorConverter().to("cuda" if torch.cuda.is_available() else "cpu")
|
| 297 |
-
|
| 298 |
-
# === Transcribe & Diarize Tab ===
|
| 299 |
whisper_model = WhisperModel("base")
|
| 300 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 301 |
def diarize_and_transcribe(audio_path):
|
|
|
|
|
|
|
|
|
|
| 302 |
# Run diarization
|
| 303 |
audio = AudioSegment.from_file(audio_path)
|
| 304 |
temp_wav = os.path.join(tempfile.gettempdir(), "diarize.wav")
|
| 305 |
audio.export(temp_wav, format="wav")
|
| 306 |
-
diarization = diarize_model(temp_wav)
|
| 307 |
-
|
| 308 |
-
# Run transcription
|
| 309 |
-
result = whisper.transcribe(temp_wav)
|
| 310 |
-
|
| 311 |
-
segments = []
|
| 312 |
-
for turn, _, speaker in diarization.itertracks(yield_label=True):
|
| 313 |
-
text = " ".join([seg.text for seg in result["segments"] if seg["start"] >= turn.start and seg["end"] <= turn.end])
|
| 314 |
-
segments.append({
|
| 315 |
-
"speaker": speaker,
|
| 316 |
-
"start": turn.start,
|
| 317 |
-
"end": turn.end,
|
| 318 |
-
"text": text
|
| 319 |
-
})
|
| 320 |
-
|
| 321 |
-
return segments
|
| 322 |
-
|
| 323 |
-
# === Voice Cloning (Dubbing) ===
|
| 324 |
-
def clone_voice(source_audio, target_audio, text):
|
| 325 |
-
source_se, _ = get_se(source_audio)
|
| 326 |
-
target_se, _ = get_se(target_audio)
|
| 327 |
-
|
| 328 |
-
out_path = os.path.join(tempfile.gettempdir(), "cloned_output.wav")
|
| 329 |
-
|
| 330 |
-
tts_model.tts_to_file(text=text, file_path=out_path)
|
| 331 |
-
tone_converter.convert(
|
| 332 |
-
audio_src_path=out_path,
|
| 333 |
-
src_se=source_se,
|
| 334 |
-
tgt_se=target_se,
|
| 335 |
-
output_path=out_path
|
| 336 |
-
)
|
| 337 |
-
return out_path
|
| 338 |
|
| 339 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 340 |
effect_options = [
|
| 341 |
"Noise Reduction",
|
| 342 |
"Compress Dynamic Range",
|
|
@@ -367,7 +317,6 @@ with gr.Blocks(title="AI Audio Studio", css="style.css") as demo:
|
|
| 367 |
gr.Audio(label="Processed Audio", type="filepath"),
|
| 368 |
gr.Image(label="Waveform Preview"),
|
| 369 |
gr.Textbox(label="Session Log (JSON)", lines=5),
|
| 370 |
-
gr.Textbox(label="Detected Genre", lines=1),
|
| 371 |
gr.Textbox(label="Status", value="β
Ready", lines=1)
|
| 372 |
],
|
| 373 |
title="Edit One File at a Time",
|
|
@@ -377,7 +326,53 @@ with gr.Blocks(title="AI Audio Studio", css="style.css") as demo:
|
|
| 377 |
clear_btn=None
|
| 378 |
)
|
| 379 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 380 |
# --- Batch Processing ---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 381 |
with gr.Tab("π Batch Processing"):
|
| 382 |
gr.Interface(
|
| 383 |
fn=batch_process_audio,
|
|
@@ -399,78 +394,4 @@ with gr.Blocks(title="AI Audio Studio", css="style.css") as demo:
|
|
| 399 |
clear_btn=None
|
| 400 |
)
|
| 401 |
|
| 402 |
-
# --- Remix Mode ---
|
| 403 |
-
with gr.Tab("π Remix Mode"):
|
| 404 |
-
gr.Interface(
|
| 405 |
-
fn=stem_split,
|
| 406 |
-
inputs=gr.Audio(label="Upload Music Track", type="filepath"),
|
| 407 |
-
outputs=[
|
| 408 |
-
gr.File(label="Vocals"),
|
| 409 |
-
gr.File(label="Drums"),
|
| 410 |
-
gr.File(label="Bass"),
|
| 411 |
-
gr.File(label="Other")
|
| 412 |
-
],
|
| 413 |
-
title="Split Into Drums, Bass, Vocals, and More",
|
| 414 |
-
description="Use AI to separate musical elements like vocals, drums, and bass.",
|
| 415 |
-
flagging_mode="never",
|
| 416 |
-
clear_btn=None
|
| 417 |
-
)
|
| 418 |
-
|
| 419 |
-
# --- Transcribe & Edit ===
|
| 420 |
-
with gr.Tab("π Transcribe & Edit"):
|
| 421 |
-
gr.Interface(
|
| 422 |
-
fn=transcribe_audio,
|
| 423 |
-
inputs=gr.Audio(label="Upload Audio", type="filepath"),
|
| 424 |
-
outputs=gr.Textbox(label="Transcribed Text", lines=10),
|
| 425 |
-
title="Transcribe & Edit Spoken Content",
|
| 426 |
-
description="Convert voice to text and edit it before exporting again."
|
| 427 |
-
)
|
| 428 |
-
|
| 429 |
-
# --- Speaker Diarization ===
|
| 430 |
-
with gr.Tab("π§ββοΈ Who Spoke When?"):
|
| 431 |
-
gr.Interface(
|
| 432 |
-
fn=diarize_and_transcribe,
|
| 433 |
-
inputs=gr.Audio(label="Upload Interview/Podcast", type="filepath"),
|
| 434 |
-
outputs=gr.JSON(label="Diarized Transcript"),
|
| 435 |
-
title="Split By Speaker + Transcribe",
|
| 436 |
-
description="Detect speakers and transcribe their speech automatically."
|
| 437 |
-
)
|
| 438 |
-
|
| 439 |
-
# --- Voice Cloning (Dubbing) ===
|
| 440 |
-
with gr.Tab("π Voice Cloning (Dubbing)"):
|
| 441 |
-
gr.Interface(
|
| 442 |
-
fn=clone_voice,
|
| 443 |
-
inputs=[
|
| 444 |
-
gr.File(label="Source Voice Clip"),
|
| 445 |
-
gr.File(label="Target Voice Clip"),
|
| 446 |
-
gr.Textbox(label="Text to Clone", lines=5)
|
| 447 |
-
],
|
| 448 |
-
outputs=gr.Audio(label="Cloned Output", type="filepath"),
|
| 449 |
-
title="Replace One Voice With Another",
|
| 450 |
-
description="Clone voice from source to target speaker using AI"
|
| 451 |
-
)
|
| 452 |
-
|
| 453 |
-
# --- TTS Voice Generator ===
|
| 454 |
-
with gr.Tab("π¬ TTS Voice Generator"):
|
| 455 |
-
gr.Interface(
|
| 456 |
-
fn=generate_tts,
|
| 457 |
-
inputs=gr.Textbox(label="Enter Text", lines=5),
|
| 458 |
-
outputs=gr.Audio(label="Generated Speech", type="filepath"),
|
| 459 |
-
title="Text-to-Speech Generator",
|
| 460 |
-
description="Type anything and turn it into natural-sounding speech."
|
| 461 |
-
)
|
| 462 |
-
|
| 463 |
-
# --- Audio Analysis Dashboard ===
|
| 464 |
-
with gr.Tab("π Audio Analysis"):
|
| 465 |
-
gr.Interface(
|
| 466 |
-
fn=analyze_audio,
|
| 467 |
-
inputs=gr.Audio(label="Upload Track", type="filepath"),
|
| 468 |
-
outputs=[
|
| 469 |
-
gr.JSON(label="Audio Stats"),
|
| 470 |
-
gr.Image(label="Waveform Graph")
|
| 471 |
-
],
|
| 472 |
-
title="View Loudness, BPM, Silence, and More",
|
| 473 |
-
description="Analyze audio loudness, tempo, and frequency content."
|
| 474 |
-
)
|
| 475 |
-
|
| 476 |
demo.launch()
|
|
|
|
| 1 |
+
import gradio as gr
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2 |
from pydub import AudioSegment
|
| 3 |
import numpy as np
|
| 4 |
import tempfile
|
| 5 |
import os
|
| 6 |
import noisereduce as nr
|
|
|
|
| 7 |
import torch
|
| 8 |
from demucs import pretrained
|
| 9 |
from demucs.apply import apply_model
|
|
|
|
| 12 |
import matplotlib.pyplot as plt
|
| 13 |
from io import BytesIO
|
| 14 |
from PIL import Image
|
| 15 |
+
import whisper
|
| 16 |
+
from faster_whisper import WhisperModel
|
| 17 |
+
import json
|
| 18 |
import datetime
|
| 19 |
import librosa
|
| 20 |
import joblib
|
| 21 |
import warnings
|
|
|
|
| 22 |
from mutagen.mp3 import MP3
|
| 23 |
from mutagen.id3 import ID3, TIT2, TPE1, TALB, TYER
|
|
|
|
|
|
|
|
|
|
|
|
|
| 24 |
|
| 25 |
# Suppress warnings
|
| 26 |
warnings.filterwarnings("ignore")
|
|
|
|
| 125 |
|
| 126 |
return [gr.File(value=path) for path in stem_paths]
|
| 127 |
|
| 128 |
+
# === Load Presets ===
|
| 129 |
def load_presets():
|
| 130 |
try:
|
| 131 |
preset_files = [f for f in os.listdir("presets") if f.endswith(".json")]
|
|
|
|
| 155 |
|
| 156 |
preset_names = list(preset_choices.keys())
|
| 157 |
|
| 158 |
+
# === Waveform Generator ===
|
| 159 |
def show_waveform(audio_file):
|
| 160 |
try:
|
| 161 |
audio = AudioSegment.from_file(audio_file)
|
|
|
|
| 171 |
except Exception as e:
|
| 172 |
return None
|
| 173 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 174 |
# === Session Info Export ===
|
| 175 |
+
def generate_session_log(audio_path, effects, isolate_vocals, export_format):
|
| 176 |
log = {
|
| 177 |
"timestamp": str(datetime.datetime.now()),
|
| 178 |
"filename": os.path.basename(audio_path),
|
| 179 |
"effects_applied": effects,
|
| 180 |
"isolate_vocals": isolate_vocals,
|
| 181 |
+
"export_format": export_format
|
|
|
|
| 182 |
}
|
| 183 |
return json.dumps(log, indent=2)
|
| 184 |
|
| 185 |
+
# === Main Processing Function ===
|
| 186 |
def process_audio(audio_file, selected_effects, isolate_vocals, preset_name, export_format):
|
| 187 |
status = "π Loading audio..."
|
| 188 |
try:
|
|
|
|
| 220 |
final_audio.export(output_path, format=export_format.lower())
|
| 221 |
|
| 222 |
waveform_image = show_waveform(output_path)
|
| 223 |
+
session_log = generate_session_log(audio_file, effects_to_apply, isolate_vocals, export_format)
|
|
|
|
| 224 |
|
| 225 |
status = "π Done!"
|
| 226 |
+
return output_path, waveform_image, session_log, status
|
| 227 |
|
| 228 |
except Exception as e:
|
| 229 |
status = f"β Error: {str(e)}"
|
| 230 |
+
return None, None, status, status
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 231 |
|
| 232 |
+
# === Transcribe & Edit Tab ===
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 233 |
whisper_model = WhisperModel("base")
|
| 234 |
|
| 235 |
+
def transcribe_audio(audio_path):
|
| 236 |
+
segments, info = whisper_model.transcribe(audio_path, beam_size=5)
|
| 237 |
+
text = " ".join([seg.text for seg in segments])
|
| 238 |
+
return text
|
| 239 |
+
|
| 240 |
+
# === Speaker Diarization Tab ===
|
| 241 |
+
try:
|
| 242 |
+
from pyannote.audio import Pipeline as DiarizationPipeline
|
| 243 |
+
from huggingface_hub import login
|
| 244 |
+
|
| 245 |
+
hf_token = os.getenv("HF_TOKEN")
|
| 246 |
+
if hf_token:
|
| 247 |
+
login(token=hf_token)
|
| 248 |
+
else:
|
| 249 |
+
print("β οΈ HF_TOKEN not set β some models may not load")
|
| 250 |
+
|
| 251 |
+
diarize_pipeline = DiarizationPipeline.from_pretrained(
|
| 252 |
+
"pyannote/speaker-diarization",
|
| 253 |
+
use_auth_token=hf_token or True
|
| 254 |
+
)
|
| 255 |
+
except Exception as e:
|
| 256 |
+
print(f"β οΈ Failed to load diarization: {e}")
|
| 257 |
+
diarize_pipeline = None
|
| 258 |
+
|
| 259 |
def diarize_and_transcribe(audio_path):
|
| 260 |
+
if diarize_pipeline is None:
|
| 261 |
+
return "β οΈ Diarization model not loaded β check HF_TOKEN"
|
| 262 |
+
|
| 263 |
# Run diarization
|
| 264 |
audio = AudioSegment.from_file(audio_path)
|
| 265 |
temp_wav = os.path.join(tempfile.gettempdir(), "diarize.wav")
|
| 266 |
audio.export(temp_wav, format="wav")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 267 |
|
| 268 |
+
try:
|
| 269 |
+
from pyannote.audio import Pipeline as DiarizationPipeline
|
| 270 |
+
diarization = diarize_pipeline(temp_wav)
|
| 271 |
+
|
| 272 |
+
# Run transcription
|
| 273 |
+
result = whisper.transcribe(temp_wav)
|
| 274 |
+
|
| 275 |
+
segments = []
|
| 276 |
+
for turn, _, speaker in diarization.itertracks(yield_label=True):
|
| 277 |
+
text = " ".join([seg["text"] for seg in result["segments"] if seg["start"] >= turn.start and seg["end"] <= turn.end])
|
| 278 |
+
segments.append({
|
| 279 |
+
"speaker": speaker,
|
| 280 |
+
"start": turn.start,
|
| 281 |
+
"end": turn.end,
|
| 282 |
+
"text": text
|
| 283 |
+
})
|
| 284 |
+
|
| 285 |
+
return segments
|
| 286 |
+
except Exception as e:
|
| 287 |
+
return f"β οΈ Diarization failed: {str(e)}"
|
| 288 |
+
|
| 289 |
+
# === UI Setup ===
|
| 290 |
effect_options = [
|
| 291 |
"Noise Reduction",
|
| 292 |
"Compress Dynamic Range",
|
|
|
|
| 317 |
gr.Audio(label="Processed Audio", type="filepath"),
|
| 318 |
gr.Image(label="Waveform Preview"),
|
| 319 |
gr.Textbox(label="Session Log (JSON)", lines=5),
|
|
|
|
| 320 |
gr.Textbox(label="Status", value="β
Ready", lines=1)
|
| 321 |
],
|
| 322 |
title="Edit One File at a Time",
|
|
|
|
| 326 |
clear_btn=None
|
| 327 |
)
|
| 328 |
|
| 329 |
+
# --- Transcribe & Edit Tab ---
|
| 330 |
+
with gr.Tab("π Transcribe & Edit"):
|
| 331 |
+
gr.Interface(
|
| 332 |
+
fn=transcribe_audio,
|
| 333 |
+
inputs=gr.Audio(label="Upload Audio", type="filepath"),
|
| 334 |
+
outputs=gr.Textbox(label="Transcribed Text", lines=10),
|
| 335 |
+
title="Transcribe Spoken Content",
|
| 336 |
+
description="Convert voice to text and edit it before exporting again."
|
| 337 |
+
)
|
| 338 |
+
|
| 339 |
+
# --- Diarization Tab (Who Spoke When?) ---
|
| 340 |
+
if diarize_pipeline:
|
| 341 |
+
with gr.Tab("π§ββοΈ Who Spoke When?"):
|
| 342 |
+
gr.Interface(
|
| 343 |
+
fn=diarize_and_transcribe,
|
| 344 |
+
inputs=gr.Audio(label="Upload Interview/Podcast", type="filepath"),
|
| 345 |
+
outputs=gr.JSON(label="Diarized Transcript"),
|
| 346 |
+
title="Split By Speaker + Transcribe",
|
| 347 |
+
description="Use AI to split podcast by speaker and transcribe their speech.",
|
| 348 |
+
flagging_mode="never"
|
| 349 |
+
)
|
| 350 |
+
|
| 351 |
# --- Batch Processing ---
|
| 352 |
+
def batch_process_audio(files, selected_effects, isolate_vocals, preset_name, export_format):
|
| 353 |
+
status = "π Loading files..."
|
| 354 |
+
try:
|
| 355 |
+
output_dir = tempfile.mkdtemp()
|
| 356 |
+
results = []
|
| 357 |
+
session_logs = []
|
| 358 |
+
|
| 359 |
+
for file in files:
|
| 360 |
+
processed_path, _, log, _ = process_audio(file.name, selected_effects, isolate_vocals, preset_name, export_format)
|
| 361 |
+
results.append(processed_path)
|
| 362 |
+
session_logs.append(log)
|
| 363 |
+
|
| 364 |
+
zip_path = os.path.join(output_dir, "batch_output.zip")
|
| 365 |
+
with zipfile.ZipFile(zip_path, 'w') as zipf:
|
| 366 |
+
for i, res in enumerate(results):
|
| 367 |
+
filename = f"processed_{i}.{export_format.lower()}"
|
| 368 |
+
zipf.write(res, filename)
|
| 369 |
+
zipf.writestr(f"session_info_{i}.json", session_logs[i])
|
| 370 |
+
|
| 371 |
+
return zip_path, "π¦ ZIP created successfully!"
|
| 372 |
+
|
| 373 |
+
except Exception as e:
|
| 374 |
+
return None, f"β Batch processing failed: {str(e)}"
|
| 375 |
+
|
| 376 |
with gr.Tab("π Batch Processing"):
|
| 377 |
gr.Interface(
|
| 378 |
fn=batch_process_audio,
|
|
|
|
| 394 |
clear_btn=None
|
| 395 |
)
|
| 396 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 397 |
demo.launch()
|