Import video pipeline as it was

1 files changed, 48 insertions, 0 deletions

diff --git a/scripts/repick.py b/scripts/repick.py
new file mode 100644
index 0000000..5f75bcf
--- /dev/null
+++ b/scripts/repick.py
@@ -0,0 +1,48 @@
+# -*- coding: utf-8 -*-
+"""Vuelve a elegir el mejor candidato ya sintetizado, comparando numeros en palabras."""
+import json, os, shutil, sys, wave
+from difflib import SequenceMatcher
+
+ROOT = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
+sys.path.insert(0, os.path.join(ROOT, "scripts"))
+from narration import SCRIPTS
+from verify import norm
+
+
+def dur(p):
+    with wave.open(p) as w:
+        return w.getnframes() / w.getframerate()
+
+
+def main(name, idxs):
+    work = os.path.join(ROOT, "tmp", name + "_rg")
+    outdir = os.path.join(ROOT, "audio", name)
+    data = json.load(open(os.path.join(outdir, "segments.json"), encoding="utf-8"))
+    segs = SCRIPTS[name]["segments"]
+    for i in idxs:
+        cands = []
+        for k in range(10):
+            w = os.path.join(work, f"{i:02d}_{k}.wav")
+            t = os.path.join(work, f"{i:02d}_{k}.txt")
+            if os.path.exists(w) and os.path.exists(t):
+                got = open(t, encoding="utf-8").read().strip()
+                r = SequenceMatcher(None, norm(segs[i][0]), norm(got)).ratio()
+                cands.append((r, k, w, got))
+        if not cands:
+            print(f"  {i:02d} sin candidatos"); continue
+        cands.sort(reverse=True)
+        for r, k, w, got in cands:
+            print(f"  {i:02d} seed{k} r={r:.2f} {dur(w):.2f}s | {got[:72]}")
+        r, k, w, got = cands[0]
+        shutil.copy(w, os.path.join(outdir, f"{i:02d}.wav"))
+        data["segments"][i].update(tts=segs[i][0], sub=segs[i][1],
+                                   dur=round(dur(w), 3), score=round(r, 3))
+        print(f"  -> {i:02d} elegido seed{k} r={r:.2f}\n")
+    data["total_audio"] = round(sum(s["dur"] for s in data["segments"]), 2)
+    json.dump(data, open(os.path.join(outdir, "segments.json"), "w", encoding="utf-8"),
+              ensure_ascii=False, indent=2)
+    print(f"[{name}] audio={data['total_audio']}s")
+
+
+if __name__ == "__main__":
+    main(sys.argv[1], [int(x) for x in sys.argv[2:]])