v5: refactor into modular stream_tools desktop app

Same inaSpeechSegmenter engine split into reusable core/ (fftools, segmenter, exporters) + utils/ and a CustomTkinter multi-tab UI. Adds JSON and ffmpeg-script export. Extracted from the stream_tools app.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

core/exporters.py

@@ -0,0 +1,188 @@
+"""
+Export singing segments to various formats:
+  - EDL (DaVinci Resolve timeline import)
+  - CSV (human-readable)
+  - DaVinci Resolve Markers CSV
+  - JSON (programmatic)
+  - ffmpeg concat script (auto-cut)
+"""
+
+import csv
+import json
+import os
+import sys
+from pathlib import Path
+
+from utils.formats import format_timecode, seconds_to_smpte
+
+
+def export_edl(segments: list, output_path: str, fps: float = 29.97,
+               title: str = "Singing Segments", source_filename: str = None):
+    """Export EDL file for DaVinci Resolve."""
+    if source_filename:
+        clip_name = source_filename
+        reel_name_safe = Path(source_filename).stem[:32].replace(" ", "_")
+    else:
+        clip_name = None
+        reel_name_safe = "001"
+
+    rec_offset = 3600.0
+
+    with open(output_path, "w", encoding="utf-8") as f:
+        f.write(f"TITLE: {title}\n")
+        f.write("FCM: NON-DROP FRAME\n\n")
+
+        current_rec_pos = rec_offset
+        for seg in segments:
+            edit_num = f"{seg['index']:03d}"
+            src_in = seconds_to_smpte(seg["start"], fps)
+            src_out = seconds_to_smpte(seg["end"], fps)
+            rec_in = seconds_to_smpte(current_rec_pos, fps)
+            rec_out = seconds_to_smpte(current_rec_pos + seg["duration"], fps)
+
+            f.write(f"{edit_num}  {reel_name_safe}  V     C        {src_in} {src_out} {rec_in} {rec_out}\n")
+
+            if clip_name:
+                f.write(f"* FROM CLIP NAME: {clip_name}\n")
+
+            f.write(f"* COMMENT: Song {seg['index']} - Duration {seg['duration']:.0f}s\n\n")
+            current_rec_pos += seg["duration"]
+
+
+def export_csv(segments: list, output_path: str):
+    """Export singing segments as CSV."""
+    with open(output_path, "w", newline="", encoding="utf-8") as f:
+        writer = csv.writer(f)
+        writer.writerow(["index", "start_sec", "end_sec", "duration_sec",
+                         "start_timecode", "end_timecode"])
+        for seg in segments:
+            writer.writerow([
+                seg["index"],
+                f"{seg['start']:.2f}",
+                f"{seg['end']:.2f}",
+                f"{seg['duration']:.1f}",
+                format_timecode(seg["start"]),
+                format_timecode(seg["end"]),
+            ])
+
+
+def export_markers_csv(segments: list, output_path: str, fps: float = 29.97):
+    """Export DaVinci Resolve Markers CSV."""
+    with open(output_path, "w", newline="", encoding="utf-8") as f:
+        writer = csv.writer(f)
+        writer.writerow(["#", "Color", "Name", "Start TC", "End TC", "Duration TC", "Notes"])
+        for seg in segments:
+            writer.writerow([
+                seg["index"],
+                "Blue",
+                f"Song {seg['index']}",
+                seconds_to_smpte(seg["start"], fps),
+                seconds_to_smpte(seg["end"], fps),
+                seconds_to_smpte(seg["duration"], fps),
+                f"Duration: {seg['duration']:.0f}s",
+            ])
+
+
+def export_json(segments: list, output_path: str):
+    """Export as JSON."""
+    with open(output_path, "w", encoding="utf-8") as f:
+        json.dump(segments, f, indent=2, ensure_ascii=False)
+
+
+def export_ffmpeg_script(segments: list, video_path: str, output_dir: str):
+    """
+    Export a script that uses ffmpeg to cut and concatenate all
+    singing segments into a single video file.
+    """
+    video_name = Path(video_path).stem
+    ext = Path(video_path).suffix
+    out_video = str(Path(output_dir) / f"{video_name}_singing_only{ext}")
+
+    is_windows = os.name == "nt" or sys.platform == "win32"
+    script_ext = ".bat" if is_windows else ".sh"
+    script_path = str(Path(output_dir) / f"{video_name}_autocut{script_ext}")
+    concat_list_path = str(Path(output_dir) / f"{video_name}_concat_list.txt")
+
+    lines = []
+    if is_windows:
+        lines.append("@echo off")
+        lines.append("REM Auto-generated ffmpeg script to extract singing segments")
+        lines.append(f'REM Source: {video_path}')
+        lines.append("")
+    else:
+        lines.append("#!/bin/bash")
+        lines.append("# Auto-generated ffmpeg script to extract singing segments")
+        lines.append(f'# Source: {video_path}')
+        lines.append("")
+
+    segment_files = []
+    for seg in segments:
+        seg_file = f"_seg_{seg['index']:03d}{ext}"
+        seg_path = str(Path(output_dir) / seg_file)
+        segment_files.append(seg_file)
+
+        cmd = (f'ffmpeg -y -ss {seg["start"]:.2f} -i "{video_path}" '
+               f'-t {seg["duration"]:.2f} -c copy "{seg_path}"')
+        lines.append(f"echo Extracting Song {seg['index']}...")
+        lines.append(cmd)
+        lines.append("")
+
+    lines.append("echo Creating concat list...")
+    if is_windows:
+        lines.append("(")
+        for sf in segment_files:
+            seg_path = str(Path(output_dir) / sf)
+            lines.append(f"  echo file '{seg_path}'")
+        lines.append(f') > "{concat_list_path}"')
+    else:
+        for i, sf in enumerate(segment_files):
+            seg_path = str(Path(output_dir) / sf)
+            op = ">" if i == 0 else ">>"
+            lines.append(f"echo \"file '{seg_path}'\" {op} \"{concat_list_path}\"")
+
+    lines.append("")
+    lines.append("echo Concatenating all segments...")
+    lines.append(f'ffmpeg -y -f concat -safe 0 -i "{concat_list_path}" -c copy "{out_video}"')
+    lines.append("")
+    lines.append("echo Cleaning up temp files...")
+
+    for sf in segment_files:
+        seg_path = str(Path(output_dir) / sf)
+        if is_windows:
+            lines.append(f'del "{seg_path}"')
+        else:
+            lines.append(f'rm -f "{seg_path}"')
+    if is_windows:
+        lines.append(f'del "{concat_list_path}"')
+    else:
+        lines.append(f'rm -f "{concat_list_path}"')
+
+    lines.append("")
+    lines.append(f'echo Done! Output: {out_video}')
+    if is_windows:
+        lines.append("pause")
+
+    with open(script_path, "w", encoding="utf-8") as f:
+        f.write("\n".join(lines))
+
+    if not is_windows:
+        os.chmod(script_path, 0o755)
+
+    return script_path
+
+
+def export_raw_segments(segments: list, output_path: str):
+    """Export ALL raw segments (speech/music/noise) as CSV for debugging."""
+    with open(output_path, "w", newline="", encoding="utf-8") as f:
+        writer = csv.writer(f)
+        writer.writerow(["label", "start_sec", "end_sec", "duration_sec",
+                         "start_timecode", "end_timecode"])
+        for label, start, end in segments:
+            writer.writerow([
+                label,
+                f"{start:.2f}",
+                f"{end:.2f}",
+                f"{end - start:.1f}",
+                format_timecode(start),
+                format_timecode(end),
+            ])

core/fftools.py

@@ -0,0 +1,99 @@
+"""
+ffmpeg / ffprobe discovery and probing helpers.
+Shared by both concat and singing detection features.
+"""
+
+import json
+import os
+import shutil
+import subprocess
+import sys
+from pathlib import Path
+
+# ── Suppress console windows on Windows ──────────────────────────────────────
+_NO_WINDOW: dict = {}
+if sys.platform == "win32":
+    _NO_WINDOW["creationflags"] = subprocess.CREATE_NO_WINDOW
+
+
+def app_dir() -> str:
+    if getattr(sys, "frozen", False):
+        return os.path.dirname(sys.executable)
+    return os.path.dirname(os.path.abspath(__file__))
+
+
+def find_tool(name: str) -> str | None:
+    """Locate ffmpeg / ffprobe next to the exe / script, then on PATH."""
+    base = app_dir()
+    for sub in ["_internal", "..", ""]:
+        for variant in [name + ".exe", name]:
+            p = os.path.join(base, sub, variant)
+            if os.path.exists(p):
+                return os.path.abspath(p)
+    return shutil.which(name + ".exe") or shutil.which(name)
+
+
+def probe_duration(path: str, ffprobe_bin: str | None = None) -> float:
+    """Get video/audio duration in seconds via ffprobe."""
+    ffprobe = ffprobe_bin or find_tool("ffprobe")
+    if not ffprobe:
+        return 0.0
+    try:
+        r = subprocess.run(
+            [ffprobe, "-v", "error",
+             "-show_entries", "format=duration",
+             "-of", "json", path],
+            stdout=subprocess.PIPE, stderr=subprocess.PIPE,
+            text=True, timeout=20,
+            **_NO_WINDOW,
+        )
+        if r.returncode != 0:
+            return 0.0
+        data = json.loads(r.stdout or "{}")
+        return max(0.0, float(data.get("format", {}).get("duration", 0) or 0))
+    except Exception:
+        return 0.0
+
+
+def extract_audio(video_path: str, output_wav: str,
+                  ffmpeg_bin: str | None = None,
+                  progress_cb=None) -> str:
+    """
+    Extract audio from video file as 16kHz mono WAV.
+    Required format for inaSpeechSegmenter.
+    
+    progress_cb: optional callable(message: str) for status updates.
+    """
+    ffmpeg = ffmpeg_bin or find_tool("ffmpeg")
+    if not ffmpeg:
+        raise FileNotFoundError("ffmpeg not found. Install ffmpeg and add to PATH.")
+
+    if progress_cb:
+        progress_cb(f"Extracting audio from: {Path(video_path).name}")
+
+    cmd = [
+        ffmpeg,
+        "-i", video_path,
+        "-vn",
+        "-acodec", "pcm_s16le",
+        "-ar", "16000",
+        "-ac", "1",
+        "-y",
+        output_wav,
+    ]
+
+    result = subprocess.run(
+        cmd,
+        capture_output=True,
+        text=True,
+        timeout=1800,
+        **_NO_WINDOW,
+    )
+    if result.returncode != 0:
+        raise RuntimeError(f"ffmpeg failed:\n{result.stderr[-500:]}")
+
+    if progress_cb:
+        size_mb = os.path.getsize(output_wav) / (1024 * 1024)
+        progress_cb(f"Audio extracted: {size_mb:.1f} MB")
+
+    return output_wav

core/segmenter.py

@@ -0,0 +1,118 @@
+"""
+Singing segment detection via inaSpeechSegmenter.
+
+Pipeline:
+  1. Run inaSpeechSegmenter to classify speech/music/noise
+  2. Filter for 'music' segments (singing voice is classified as music)
+  3. Merge nearby segments, apply padding, filter by minimum duration
+"""
+
+import time
+from utils.formats import format_timecode
+
+
+def check_segmenter_available() -> tuple[bool, str]:
+    """Check if inaSpeechSegmenter is installed."""
+    try:
+        from inaSpeechSegmenter import Segmenter  # noqa: F401
+        return True, ""
+    except ImportError:
+        return False, (
+            "inaSpeechSegmenter is not installed.\n\n"
+            "Install it with:\n"
+            "  pip install inaSpeechSegmenter\n\n"
+            "You also need TensorFlow:\n"
+            "  pip install tensorflow[and-cuda]   (GPU)\n"
+            "  pip install tensorflow              (CPU)"
+        )
+
+
+def run_segmentation(wav_path: str, progress_cb=None) -> list:
+    """
+    Run inaSpeechSegmenter on audio file.
+    Returns list of (label, start_sec, end_sec) tuples.
+    Labels: 'music', 'speech', 'noise', 'noEnergy'
+    """
+    from inaSpeechSegmenter import Segmenter
+
+    if progress_cb:
+        progress_cb("Loading segmentation model…")
+
+    start_time = time.time()
+    seg = Segmenter(vad_engine='smn', detect_gender=False)
+
+    if progress_cb:
+        progress_cb("Running audio segmentation (this may take a while)…")
+
+    segments = seg(wav_path)
+    elapsed = time.time() - start_time
+
+    if progress_cb:
+        progress_cb(f"Segmentation complete in {elapsed:.1f}s — {len(segments)} raw segments")
+
+    return segments
+
+
+def merge_singing_segments(
+    segments: list,
+    max_gap: float = 10.0,
+    min_duration: float = 20.0,
+    padding: float = 2.0,
+) -> list:
+    """
+    Filter for 'music' segments and merge nearby ones.
+    
+    Returns list of dicts:
+        [{"index", "start", "end", "duration", "original_start", "original_end"}, ...]
+    """
+    music_segs = [(start, end) for label, start, end in segments if label == "music"]
+
+    if not music_segs:
+        return []
+
+    music_segs.sort(key=lambda x: x[0])
+
+    # Merge segments with gaps smaller than max_gap
+    merged = []
+    current_start, current_end = music_segs[0]
+
+    for start, end in music_segs[1:]:
+        if start - current_end <= max_gap:
+            current_end = max(current_end, end)
+        else:
+            merged.append((current_start, current_end))
+            current_start, current_end = start, end
+    merged.append((current_start, current_end))
+
+    # Apply padding and minimum duration filter
+    results = []
+    idx = 1
+    for start, end in merged:
+        duration = end - start
+        if duration >= min_duration:
+            padded_start = max(0, start - padding)
+            padded_end = end + padding
+            results.append({
+                "index": idx,
+                "start": padded_start,
+                "end": padded_end,
+                "duration": padded_end - padded_start,
+                "original_start": start,
+                "original_end": end,
+            })
+            idx += 1
+
+    return results
+
+
+def get_segment_stats(segments: list) -> dict:
+    """Get summary statistics from raw segmentation output."""
+    type_counts = {}
+    type_durations = {}
+    for label, start, end in segments:
+        type_counts[label] = type_counts.get(label, 0) + 1
+        type_durations[label] = type_durations.get(label, 0) + (end - start)
+    return {
+        "counts": type_counts,
+        "durations": type_durations,
+    }

sing_detect.py

@@ -1,108 +0,0 @@
-import subprocess
-import os
-import torch
-import librosa
-import numpy as np
-from huggingface_hub import hf_hub_download
-import tkinter as tk
-from tkinter import filedialog
-
-# ------------------------
-# 1. Extract audio from video
-# ------------------------
-def extract_audio(input_video, output_wav):
-    cmd = [
-        "ffmpeg", "-y",
-        "-i", input_video,
-        "-vn",          # no video
-        "-ac", "1",     # mono
-        "-ar", "16000", # 16kHz
-        output_wav
-    ]
-    subprocess.run(cmd, check=True)
-
-# ------------------------
-# 2. Load pretrained model
-# ------------------------
-def load_model():
-    repo_id = "matthijs/svd"  # HuggingFace singing voice detection model
-    filename = hf_hub_download(repo_id=repo_id, filename="model.pt")
-    model = torch.jit.load(filename)
-    model.eval()
-    return model
-
-# ------------------------
-# 3. Run inference on audio
-# ------------------------
-def detect_singing(audio_path, model, hop_length=512, threshold=0.5, min_duration=10.0):
-    y, sr = librosa.load(audio_path, sr=16000)
-    x = torch.tensor(y).float().unsqueeze(0)
-
-    with torch.no_grad():
-        pred = model(x).squeeze().numpy()
-
-    times = librosa.frames_to_time(np.arange(len(pred)), sr=sr, hop_length=hop_length)
-
-    segments = []
-    in_segment = False
-    seg_start = None
-
-    for t, p in zip(times, pred):
-        if p > threshold and not in_segment:
-            in_segment = True
-            seg_start = t
-        elif p <= threshold and in_segment:
-            in_segment = False
-            seg_end = t
-            if seg_end - seg_start >= min_duration:
-                segments.append((seg_start, seg_end))
-
-    if in_segment:
-        seg_end = times[-1]
-        if seg_end - seg_start >= min_duration:
-            segments.append((seg_start, seg_end))
-
-    return segments
-
-# ------------------------
-# 4. Export CSV
-# ------------------------
-def export_csv(segments, out_file="segments.csv"):
-    with open(out_file, "w", encoding="utf-8") as f:
-        f.write("start_time,end_time\n")
-        for s, e in segments:
-            f.write(f"{s:.2f},{e:.2f}\n")
-    print(f"[+] Saved: {out_file}")
-
-# ------------------------
-# MAIN
-# ------------------------
-if __name__ == "__main__":
-    # Tkinter GUI to choose file
-    root = tk.Tk()
-    root.withdraw()
-    input_video = filedialog.askopenfilename(
-        title="Choose your video file",
-        filetypes=[("Video files", "*.mp4 *.mkv *.flv *.mov *.avi"), ("All files", "*.*")]
-    )
-    if not input_video:
-        print("No file selected. Exiting.")
-        exit()
-
-    audio_file = "audio.wav"
-
-    print(f"[+] Processing: {input_video}")
-
-    # Step 1: extract audio
-    extract_audio(input_video, audio_file)
-
-    # Step 2: load model
-    model = load_model()
-
-    # Step 3: run detection
-    segments = detect_singing(audio_file, model)
-
-    # Step 4: save results
-    export_csv(segments, "segments.csv")
-
-    print("[+] Done! Detected singing segments have been saved to segments.csv")

ui/tab_detect.py

@@ -0,0 +1,505 @@
+"""
+Tab 2: Song Detection
+Detect singing segments in a video and export EDL/CSV/JSON for DaVinci Resolve.
+"""
+
+import os
+import threading
+import tkinter as tk
+from pathlib import Path
+from tkinter import filedialog, messagebox
+
+import customtkinter as ctk
+
+from core.fftools import find_tool, extract_audio
+from core.segmenter import check_segmenter_available, run_segmentation, merge_singing_segments, get_segment_stats
+from core.exporters import export_edl, export_csv, export_markers_csv, export_json, export_ffmpeg_script
+from utils.formats import format_timecode, fmt_dur, fmt_elapsed
+from ui.theme import BTN_NEUTRAL, ACCENT_PURPLE, ACCENT_GREEN
+
+
+class DetectTab:
+    """Song detection tab."""
+
+    VIDEO_EXTS = {".mp4", ".mov", ".mkv", ".avi", ".flv",
+                  ".ts", ".wmv", ".m4v", ".webm", ".mts", ".m2ts"}
+
+    def __init__(self, parent_frame, app):
+        self.frame = parent_frame
+        self.app = app
+        self.ffmpeg = app.ffmpeg
+
+        # State
+        self.input_path_var = tk.StringVar(value="")
+        self.output_dir_var = tk.StringVar(value="")
+        self.gap_var = tk.DoubleVar(value=10.0)
+        self.min_dur_var = tk.DoubleVar(value=20.0)
+        self.padding_var = tk.DoubleVar(value=2.0)
+        self.fps_var = tk.DoubleVar(value=29.97)
+        self.autocut_var = tk.BooleanVar(value=False)
+
+        self._running = False
+        self._stop_req = False
+
+        # Results
+        self._singing_segments = []
+        self._raw_segments = []
+
+        self._build_ui()
+        self._check_deps()
+
+    # ══════════════════════════════════════════════════════════════════════════
+    #  UI
+    # ══════════════════════════════════════════════════════════════════════════
+
+    def _build_ui(self):
+        scroll = ctk.CTkScrollableFrame(self.frame, corner_radius=12)
+        scroll.pack(fill="both", expand=True, padx=4, pady=4)
+
+        self._build_sec_input(scroll)
+        self._build_sec_params(scroll)
+        self._build_sec_run(scroll)
+        self._build_sec_results(scroll)
+
+    def _section(self, parent, title):
+        f = ctk.CTkFrame(parent, corner_radius=12)
+        hdr = ctk.CTkFrame(f, fg_color="transparent")
+        hdr.pack(fill="x", padx=14, pady=(10, 0))
+        ctk.CTkLabel(hdr, text=title,
+                     font=ctk.CTkFont(size=13, weight="bold")).pack(side="left")
+        ctk.CTkFrame(f, height=1, fg_color=("gray78", "gray32")).pack(
+            fill="x", padx=14, pady=(6, 0))
+        return f
+
+    # ── Input ─────────────────────────────────────────────────────────────────
+
+    def _build_sec_input(self, parent):
+        sec = self._section(parent, "1 · Input Video")
+        sec.pack(fill="x", pady=(0, 12))
+
+        body = ctk.CTkFrame(sec, fg_color="transparent")
+        body.pack(fill="x", padx=14, pady=(10, 4))
+
+        r1 = ctk.CTkFrame(body, fg_color="transparent")
+        r1.pack(fill="x", pady=(0, 4))
+        ctk.CTkLabel(r1, text="Video file:", width=96, anchor="w").pack(side="left")
+        ctk.CTkEntry(r1, textvariable=self.input_path_var).pack(side="left", fill="x", expand=True)
+        ctk.CTkButton(r1, text="Browse…", width=100, command=self._pick_input,
+                      **BTN_NEUTRAL).pack(side="left", padx=(8, 0))
+
+        # "Use Concat Output" button
+        self.use_concat_btn = ctk.CTkButton(
+            body, text="📎  Use Concat Output", width=200,
+            fg_color=ACCENT_PURPLE["fg"], text_color=ACCENT_PURPLE["text"],
+            hover_color=ACCENT_PURPLE["hover"],
+            command=self._use_concat_output)
+        self.use_concat_btn.pack(anchor="w", pady=(4, 4))
+
+        r2 = ctk.CTkFrame(body, fg_color="transparent")
+        r2.pack(fill="x", pady=(4, 0))
+        ctk.CTkLabel(r2, text="Output dir:", width=96, anchor="w").pack(side="left")
+        ctk.CTkEntry(r2, textvariable=self.output_dir_var).pack(side="left", fill="x", expand=True)
+        ctk.CTkButton(r2, text="Browse…", width=100, command=self._pick_output_dir,
+                      **BTN_NEUTRAL).pack(side="left", padx=(8, 0))
+
+        ctk.CTkLabel(sec, text="Output defaults to same directory as input video.",
+                     font=ctk.CTkFont(size=11), text_color=("gray50", "gray55")
+                     ).pack(anchor="w", padx=14, pady=(4, 10))
+
+    # ── Parameters ────────────────────────────────────────────────────────────
+
+    def _build_sec_params(self, parent):
+        sec = self._section(parent, "2 · Detection Parameters")
+        sec.pack(fill="x", pady=(0, 12))
+
+        body = ctk.CTkFrame(sec, fg_color="transparent")
+        body.pack(fill="x", padx=14, pady=(10, 10))
+
+        params = [
+            ("Merge gap (s):",     self.gap_var,     1, 60,  "Max silence between song parts to merge"),
+            ("Min duration (s):",  self.min_dur_var,  5, 120, "Ignore segments shorter than this"),
+            ("Padding (s):",       self.padding_var,  0, 15,  "Extra seconds before/after each song"),
+            ("FPS:",               self.fps_var,      23, 60, "Video frame rate for timecodes"),
+        ]
+
+        for label_text, var, lo, hi, tooltip in params:
+            row = ctk.CTkFrame(body, fg_color="transparent")
+            row.pack(fill="x", pady=3)
+
+            ctk.CTkLabel(row, text=label_text, width=130, anchor="w",
+                         font=ctk.CTkFont(size=12)).pack(side="left")
+
+            slider = ctk.CTkSlider(row, from_=lo, to=hi, variable=var,
+                                    width=200, number_of_steps=max(1, hi - lo))
+            slider.pack(side="left", padx=(0, 8))
+
+            val_lbl = ctk.CTkLabel(row, text=f"{var.get():.1f}", width=50, anchor="w",
+                                    font=ctk.CTkFont(size=12, weight="bold"))
+            val_lbl.pack(side="left")
+
+            ctk.CTkLabel(row, text=tooltip, font=ctk.CTkFont(size=11),
+                         text_color=("gray50", "gray55")).pack(side="left", padx=(10, 0))
+
+            # Update label on slider move
+            var.trace_add("write", lambda *_, v=var, l=val_lbl: l.configure(text=f"{v.get():.1f}"))
+
+        # Auto-cut checkbox
+        ctk.CTkCheckBox(body, text="Generate auto-cut ffmpeg script (no DaVinci needed)",
+                        variable=self.autocut_var).pack(anchor="w", pady=(8, 0))
+
+    # ── Run ───────────────────────────────────────────────────────────────────
+
+    def _build_sec_run(self, parent):
+        sec = self._section(parent, "3 · Detect")
+        sec.pack(fill="x", pady=(0, 12))
+
+        body = ctk.CTkFrame(sec, fg_color="transparent")
+        body.pack(fill="x", padx=14, pady=(12, 6))
+
+        btn_row = ctk.CTkFrame(body, fg_color="transparent")
+        btn_row.pack(fill="x", pady=(0, 10))
+
+        self.detect_btn = ctk.CTkButton(
+            btn_row, text="🎤  Start Detection", width=230, height=42,
+            font=ctk.CTkFont(size=14, weight="bold"),
+            fg_color=ACCENT_GREEN["fg"], hover_color=ACCENT_GREEN["hover"],
+            command=self.start_detect)
+        self.detect_btn.pack(side="left")
+
+        self.stop_btn = ctk.CTkButton(
+            btn_row, text="■  Stop", width=106, height=42,
+            fg_color=("gray80", "gray25"), text_color=("gray10", "gray90"),
+            hover_color=("#FCA5A5", "#7F1D1D"),
+            command=self._stop, state="disabled")
+        self.stop_btn.pack(side="left", padx=(12, 0))
+
+        # Dependency warning
+        self.dep_lbl = ctk.CTkLabel(body, text="", font=ctk.CTkFont(size=11),
+                                     text_color="#F97316", wraplength=600, anchor="w", justify="left")
+        self.dep_lbl.pack(fill="x", pady=(0, 4))
+
+        # Progress
+        self.progress = ctk.CTkProgressBar(body, height=16, corner_radius=8, mode="indeterminate")
+        self.progress.pack(fill="x", pady=(0, 6))
+        self.progress.stop()
+        self.progress.set(0)
+
+        self.status_lbl = ctk.CTkLabel(body, text="", anchor="w", font=ctk.CTkFont(size=12))
+        self.status_lbl.pack(fill="x")
+
+        ctk.CTkFrame(sec, height=6, fg_color="transparent").pack()
+
+    # ── Results ───────────────────────────────────────────────────────────────
+
+    def _build_sec_results(self, parent):
+        sec = self._section(parent, "4 · Results")
+        sec.pack(fill="x", pady=(0, 14))
+
+        self.results_body = ctk.CTkFrame(sec, fg_color="transparent")
+        self.results_body.pack(fill="x", padx=14, pady=(10, 10))
+
+        self.results_lbl = ctk.CTkLabel(
+            self.results_body, text="No results yet. Run detection first.",
+            font=ctk.CTkFont(size=12), text_color=("gray50", "gray55"))
+        self.results_lbl.pack(anchor="w")
+
+        # Song list (populated after detection)
+        self.song_list_frame = ctk.CTkFrame(self.results_body, fg_color="transparent")
+
+        # Export buttons (hidden until results)
+        self.export_frame = ctk.CTkFrame(sec, fg_color="transparent")
+
+    # ══════════════════════════════════════════════════════════════════════════
+    #  Logic
+    # ══════════════════════════════════════════════════════════════════════════
+
+    def _check_deps(self):
+        available, msg = check_segmenter_available()
+        if not available:
+            self.dep_lbl.configure(text=f"⚠  {msg}")
+            self.detect_btn.configure(state="disabled")
+        else:
+            self.dep_lbl.configure(text="")
+
+        if not self.ffmpeg:
+            self.dep_lbl.configure(text="⚠  ffmpeg not found. Install ffmpeg and add to PATH.")
+            self.detect_btn.configure(state="disabled")
+
+    def _pick_input(self):
+        f = filedialog.askopenfilename(
+            title="Select video file",
+            filetypes=[("Video files", "*.mp4 *.mov *.mkv *.avi *.flv *.ts *.wmv *.m4v *.webm *.mts"),
+                       ("All files", "*.*")])
+        if f:
+            self.input_path_var.set(f)
+            if not self.output_dir_var.get().strip():
+                self.output_dir_var.set(str(Path(f).parent))
+
+    def _pick_output_dir(self):
+        d = filedialog.askdirectory(title="Choose output folder")
+        if d:
+            self.output_dir_var.set(d)
+
+    def _use_concat_output(self):
+        """Fill input from the Concat tab's output path."""
+        if hasattr(self.app, 'concat_tab'):
+            path = self.app.concat_tab.get_last_output_path()
+            if path and Path(path).exists():
+                self.input_path_var.set(path)
+                if not self.output_dir_var.get().strip():
+                    self.output_dir_var.set(str(Path(path).parent))
+                self._set_status(f"Loaded concat output: {Path(path).name}")
+            elif path:
+                self.input_path_var.set(path)
+                if not self.output_dir_var.get().strip():
+                    self.output_dir_var.set(str(Path(path).parent))
+                self._set_status("Concat output path set (file not yet created — run concat first)")
+            else:
+                messagebox.showinfo("No output", "Set an output path in the Concat tab first.")
+
+    def _set_status(self, text, error=False):
+        color = "#DC2626" if error else ("gray10", "gray90")
+        self.status_lbl.configure(text=text, text_color=color)
+
+    def _stop(self):
+        self._stop_req = True
+        self._set_status("Stopping…")
+
+    def start_detect(self):
+        """Start the detection pipeline in a background thread."""
+        input_path = self.input_path_var.get().strip()
+        if not input_path:
+            messagebox.showerror("No input", "Select a video file first.")
+            return
+        if not os.path.isfile(input_path):
+            messagebox.showerror("File not found", f"Cannot find:\n{input_path}")
+            return
+
+        available, msg = check_segmenter_available()
+        if not available:
+            messagebox.showerror("Missing dependency", msg)
+            return
+
+        self._running = True
+        self._stop_req = False
+        self.detect_btn.configure(state="disabled")
+        self.stop_btn.configure(state="normal")
+        self.progress.configure(mode="indeterminate")
+        self.progress.start()
+        self._set_status("Starting detection…")
+
+        threading.Thread(target=self._detect_worker, args=(input_path,), daemon=True).start()
+
+    def _detect_worker(self, input_path):
+        output_dir = self.output_dir_var.get().strip() or str(Path(input_path).parent)
+        os.makedirs(output_dir, exist_ok=True)
+        stem = Path(input_path).stem
+
+        wav_path = os.path.join(output_dir, f"{stem}_audio.wav")
+
+        try:
+            # Step 1: Extract audio
+            def progress_cb(msg):
+                self.app.after(0, lambda: self._set_status(msg))
+
+            extract_audio(input_path, wav_path, self.ffmpeg, progress_cb=progress_cb)
+
+            if self._stop_req:
+                self._cleanup_and_finish(wav_path, "Stopped by user.")
+                return
+
+            # Step 2: Run segmentation
+            self.app.after(0, lambda: self._set_status("Running audio segmentation (this may take a while)…"))
+            raw_segments = run_segmentation(wav_path, progress_cb=progress_cb)
+
+            if self._stop_req:
+                self._cleanup_and_finish(wav_path, "Stopped by user.")
+                return
+
+            self._raw_segments = raw_segments
+
+            # Step 3: Merge
+            self.app.after(0, lambda: self._set_status("Merging singing segments…"))
+            singing = merge_singing_segments(
+                raw_segments,
+                max_gap=self.gap_var.get(),
+                min_duration=self.min_dur_var.get(),
+                padding=self.padding_var.get(),
+            )
+            self._singing_segments = singing
+
+            # Step 4: Export
+            if singing:
+                fps = self.fps_var.get()
+                source_name = Path(input_path).name
+
+                edl_path = os.path.join(output_dir, f"{stem}_singing.edl")
+                csv_path = os.path.join(output_dir, f"{stem}_singing.csv")
+                markers_path = os.path.join(output_dir, f"{stem}_markers.csv")
+                json_path = os.path.join(output_dir, f"{stem}_singing.json")
+
+                export_edl(singing, edl_path, fps=fps,
+                          title=f"{stem} - Singing", source_filename=source_name)
+                export_csv(singing, csv_path)
+                export_markers_csv(singing, markers_path, fps=fps)
+                export_json(singing, json_path)
+
+                if self.autocut_var.get():
+                    export_ffmpeg_script(singing, input_path, output_dir)
+
+                total_singing = sum(s["duration"] for s in singing)
+                result_msg = (f"✓  Found {len(singing)} songs  ·  "
+                              f"Total singing: {format_timecode(total_singing)}  ·  "
+                              f"Exported to: {output_dir}")
+
+                self.app.after(0, lambda: self._show_results(singing, edl_path, output_dir))
+            else:
+                result_msg = "No singing segments detected. Try lowering min duration or increasing gap."
+
+            # Cleanup WAV
+            if os.path.exists(wav_path):
+                os.remove(wav_path)
+
+            self.app.after(0, lambda: self._set_status(result_msg))
+
+        except Exception as e:
+            self.app.after(0, lambda: self._set_status(f"Error: {e}", error=True))
+            if os.path.exists(wav_path):
+                try:
+                    os.remove(wav_path)
+                except Exception:
+                    pass
+
+        self.app.after(0, self._finish_detect)
+
+    def _cleanup_and_finish(self, wav_path, msg):
+        if os.path.exists(wav_path):
+            try:
+                os.remove(wav_path)
+            except Exception:
+                pass
+        self.app.after(0, lambda: self._set_status(msg))
+        self.app.after(0, self._finish_detect)
+
+    def _finish_detect(self):
+        self._running = False
+        self._stop_req = False
+        self.detect_btn.configure(state="normal")
+        self.stop_btn.configure(state="disabled")
+        self.progress.stop()
+        self.progress.configure(mode="determinate")
+        self.progress.set(1.0 if self._singing_segments else 0)
+
+    def _show_results(self, segments, edl_path, output_dir):
+        """Display detection results in the Results section."""
+        # Clear previous
+        for w in self.song_list_frame.winfo_children():
+            w.destroy()
+        self.export_frame.pack_forget()
+
+        self.results_lbl.configure(
+            text=f"Found {len(segments)} singing segments:",
+            text_color=("gray10", "gray90"))
+
+        self.song_list_frame.pack(fill="x", pady=(8, 0))
+
+        # Header
+        hdr = ctk.CTkFrame(self.song_list_frame, fg_color=("gray88", "gray20"), corner_radius=8)
+        hdr.pack(fill="x", pady=(0, 4))
+        for text, w in [("#", 40), ("Start", 90), ("End", 90), ("Duration", 80)]:
+            ctk.CTkLabel(hdr, text=text, width=w, anchor="center",
+                         font=ctk.CTkFont(size=11, weight="bold")).pack(side="left", padx=4, pady=4)
+
+        # Rows
+        for seg in segments:
+            row = ctk.CTkFrame(self.song_list_frame, corner_radius=6, height=30)
+            row.pack(fill="x", pady=2)
+            row.pack_propagate(False)
+
+            vals = [
+                (f"Song {seg['index']}", 40),
+                (format_timecode(seg["start"]), 90),
+                (format_timecode(seg["end"]), 90),
+                (f"{seg['duration']:.0f}s", 80),
+            ]
+            for text, w in vals:
+                ctk.CTkLabel(row, text=text, width=w, anchor="center",
+                             font=ctk.CTkFont(size=11)).pack(side="left", padx=4)
+
+        # Total
+        total = sum(s["duration"] for s in segments)
+        ctk.CTkLabel(self.song_list_frame,
+                     text=f"Total singing time: {format_timecode(total)}",
+                     font=ctk.CTkFont(size=12, weight="bold")).pack(anchor="w", pady=(8, 0))
+
+        # Export info
+        self.export_frame.pack(fill="x", padx=14, pady=(4, 10))
+        for w in self.export_frame.winfo_children():
+            w.destroy()
+
+        ctk.CTkLabel(self.export_frame,
+                     text=f"Files exported to: {output_dir}",
+                     font=ctk.CTkFont(size=11), text_color=("gray45", "gray55"),
+                     wraplength=600, anchor="w", justify="left").pack(anchor="w")
+
+        info_lines = [
+            "• _singing.edl  →  DaVinci Resolve: File → Import → Timeline",
+            "• _markers.csv  →  DaVinci Resolve: Timeline → Import Markers from CSV",
+            "• _singing.csv  →  Human-readable segment list",
+            "• _singing.json →  Programmatic use",
+        ]
+        if self.autocut_var.get():
+            info_lines.append("• _autocut.bat  →  Run to auto-cut singing segments (no DaVinci needed)")
+
+        for line in info_lines:
+            ctk.CTkLabel(self.export_frame, text=line,
+                         font=ctk.CTkFont(size=11),
+                         text_color=("gray50", "gray55"), anchor="w").pack(anchor="w")
+
+    # ── Public API for Quick Flow ─────────────────────────────────────────────
+
+    def run_detect_headless(self, input_path: str, output_dir: str,
+                            gap: float, min_dur: float, padding: float,
+                            fps: float, autocut: bool,
+                            progress_cb=None) -> list:
+        """
+        Run detection synchronously (called from Quick Flow worker thread).
+        Returns list of singing segments.
+        """
+        os.makedirs(output_dir, exist_ok=True)
+        stem = Path(input_path).stem
+        wav_path = os.path.join(output_dir, f"{stem}_audio.wav")
+
+        try:
+            extract_audio(input_path, wav_path, self.ffmpeg, progress_cb=progress_cb)
+
+            raw_segments = run_segmentation(wav_path, progress_cb=progress_cb)
+
+            if progress_cb:
+                progress_cb("Merging singing segments…")
+
+            singing = merge_singing_segments(raw_segments,
+                                             max_gap=gap, min_duration=min_dur, padding=padding)
+
+            if singing:
+                source_name = Path(input_path).name
+                export_edl(singing, os.path.join(output_dir, f"{stem}_singing.edl"),
+                          fps=fps, title=f"{stem} - Singing", source_filename=source_name)
+                export_csv(singing, os.path.join(output_dir, f"{stem}_singing.csv"))
+                export_markers_csv(singing, os.path.join(output_dir, f"{stem}_markers.csv"), fps=fps)
+                export_json(singing, os.path.join(output_dir, f"{stem}_singing.json"))
+                if autocut:
+                    export_ffmpeg_script(singing, input_path, output_dir)
+
+            if os.path.exists(wav_path):
+                os.remove(wav_path)
+
+            return singing
+
+        except Exception:
+            if os.path.exists(wav_path):
+                try:
+                    os.remove(wav_path)
+                except Exception:
+                    pass
+            raise

ui/theme.py

@@ -0,0 +1,64 @@
+"""
+Theme, DPI, and shared styling constants.
+"""
+
+import sys
+
+# ── DPI awareness (call before any Tk/CTk window) ────────────────────────────
+def setup_dpi():
+    if sys.platform == "win32":
+        try:
+            import ctypes
+            ctypes.windll.shcore.SetProcessDpiAwareness(2)
+        except Exception:
+            try:
+                import ctypes
+                ctypes.windll.user32.SetProcessDPIAware()
+            except Exception:
+                pass
+
+
+def dpi_scale(widget) -> float:
+    try:
+        return max(1.0, widget.winfo_fpixels("1i") / 96.0)
+    except Exception:
+        return 1.0
+
+
+# ── Color accents ─────────────────────────────────────────────────────────────
+# Each feature area has its own accent color scheme: (light_mode, dark_mode)
+
+ACCENT_BLUE = {
+    "fg":    ("#3B82F6", "#2563EB"),
+    "hover": ("#2563EB", "#1D4ED8"),
+}
+
+ACCENT_PURPLE = {
+    "fg":       ("#EDE9FE", "#3B1F6E"),
+    "text":     ("#5B21B6", "#C4B5FD"),
+    "hover":    ("#DDD6FE", "#4C2889"),
+    "preview":  ("#7C3AED", "#A78BFA"),
+}
+
+ACCENT_GREEN = {
+    "fg":    ("#10B981", "#059669"),
+    "hover": ("#059669", "#047857"),
+}
+
+ACCENT_AMBER = {
+    "fg":    ("#F59E0B", "#D97706"),
+    "hover": ("#D97706", "#B45309"),
+}
+
+# Neutral button style
+BTN_NEUTRAL = {
+    "fg_color":    ("gray80", "gray25"),
+    "text_color":  ("gray10", "gray90"),
+    "hover_color": ("gray70", "gray35"),
+}
+
+BTN_SIDE = {
+    "fg_color":    ("gray82", "gray22"),
+    "text_color":  ("gray10", "gray90"),
+    "hover_color": ("gray72", "gray32"),
+}

utils/formats.py

@@ -0,0 +1,42 @@
+"""
+Timecode & duration formatting utilities.
+"""
+
+
+def fmt_dur(seconds: float) -> str:
+    """Format a duration as H:MM:SS or M:SS for display."""
+    if seconds < 0:
+        return "…"
+    if seconds == 0:
+        return "—"
+    s = int(round(seconds))
+    h, rem = divmod(s, 3600)
+    m, sec = divmod(rem, 60)
+    return f"{h}:{m:02d}:{sec:02d}" if h else f"{m}:{sec:02d}"
+
+
+def fmt_elapsed(s: float) -> str:
+    if s < 60:
+        return f"{int(s)}s"
+    m, sec = divmod(int(s), 60)
+    return f"{m}m {sec:02d}s"
+
+
+def format_timecode(seconds: float) -> str:
+    """Format seconds as HH:MM:SS."""
+    h = int(seconds // 3600)
+    m = int((seconds % 3600) // 60)
+    s = int(seconds % 60)
+    return f"{h:02d}:{m:02d}:{s:02d}"
+
+
+def seconds_to_smpte(seconds: float, fps: float = 29.97) -> str:
+    """Convert seconds to SMPTE timecode HH:MM:SS:FF."""
+    total_frames = int(seconds * fps)
+    ff = total_frames % int(round(fps))
+    total_seconds = total_frames // int(round(fps))
+    ss = total_seconds % 60
+    total_minutes = total_seconds // 60
+    mm = total_minutes % 60
+    hh = total_minutes // 60
+    return f"{hh:02d}:{mm:02d}:{ss:02d}:{ff:02d}"