Add pipeline benchmarking script

data/mix/mixture_map.csv

@@ -0,0 +1,51 @@
+mix_id,mix_file,src1,src2
+m01,m01.wav,speaker10\speaker10_17.wav,speaker1\speaker1_16.wav
+m02,m02.wav,speaker10\speaker10_11.wav,speaker8\speaker8_08.wav
+m03,m03.wav,speaker10\speaker10_16.wav,speaker3\speaker3_06.wav
+m04,m04.wav,speaker3\speaker3_03.wav,speaker8\speaker8_16.wav
+m05,m05.wav,speaker1\speaker1_03.wav,speaker2\speaker2_19.wav
+m06,m06.wav,speaker5\speaker5_11.wav,speaker10\speaker10_20.wav
+m07,m07.wav,speaker5\speaker5_10.wav,speaker4\speaker4_04.wav
+m08,m08.wav,speaker6\speaker6_07.wav,speaker10\speaker10_18.wav
+m09,m09.wav,speaker5\speaker5_11.wav,speaker3\speaker3_10.wav
+m10,m10.wav,speaker4\speaker4_16.wav,speaker10\speaker10_12.wav
+m11,m11.wav,speaker5\speaker5_02.wav,speaker2\speaker2_02.wav
+m12,m12.wav,speaker10\speaker10_07.wav,speaker2\speaker2_16.wav
+m13,m13.wav,speaker4\speaker4_16.wav,speaker8\speaker8_19.wav
+m14,m14.wav,speaker1\speaker1_10.wav,speaker3\speaker3_19.wav
+m15,m15.wav,speaker3\speaker3_15.wav,speaker5\speaker5_05.wav
+m16,m16.wav,speaker7\speaker7_17.wav,speaker2\speaker2_13.wav
+m17,m17.wav,speaker8\speaker8_08.wav,speaker4\speaker4_03.wav
+m18,m18.wav,speaker5\speaker5_13.wav,speaker3\speaker3_06.wav
+m19,m19.wav,speaker1\speaker1_13.wav,speaker10\speaker10_14.wav
+m20,m20.wav,speaker9\speaker9_20.wav,speaker10\speaker10_20.wav
+m21,m21.wav,speaker8\speaker8_07.wav,speaker4\speaker4_01.wav
+m22,m22.wav,speaker8\speaker8_19.wav,speaker4\speaker4_12.wav
+m23,m23.wav,speaker7\speaker7_17.wav,speaker1\speaker1_06.wav
+m24,m24.wav,speaker10\speaker10_06.wav,speaker4\speaker4_09.wav
+m25,m25.wav,speaker2\speaker2_06.wav,speaker8\speaker8_01.wav
+m26,m26.wav,speaker1\speaker1_02.wav,speaker10\speaker10_18.wav
+m27,m27.wav,speaker9\speaker9_11.wav,speaker10\speaker10_05.wav
+m28,m28.wav,speaker8\speaker8_13.wav,speaker2\speaker2_05.wav
+m29,m29.wav,speaker4\speaker4_09.wav,speaker8\speaker8_03.wav
+m30,m30.wav,speaker8\speaker8_18.wav,speaker3\speaker3_20.wav
+m31,m31.wav,speaker5\speaker5_06.wav,speaker7\speaker7_04.wav
+m32,m32.wav,speaker10\speaker10_01.wav,speaker5\speaker5_08.wav
+m33,m33.wav,speaker3\speaker3_11.wav,speaker1\speaker1_10.wav
+m34,m34.wav,speaker1\speaker1_11.wav,speaker5\speaker5_06.wav
+m35,m35.wav,speaker7\speaker7_07.wav,speaker3\speaker3_13.wav
+m36,m36.wav,speaker9\speaker9_16.wav,speaker7\speaker7_05.wav
+m37,m37.wav,speaker10\speaker10_03.wav,speaker9\speaker9_02.wav
+m38,m38.wav,speaker1\speaker1_10.wav,speaker10\speaker10_20.wav
+m39,m39.wav,speaker10\speaker10_15.wav,speaker3\speaker3_15.wav
+m40,m40.wav,speaker6\speaker6_17.wav,speaker2\speaker2_04.wav
+m41,m41.wav,speaker10\speaker10_07.wav,speaker7\speaker7_12.wav
+m42,m42.wav,speaker1\speaker1_16.wav,speaker10\speaker10_14.wav
+m43,m43.wav,speaker3\speaker3_10.wav,speaker6\speaker6_14.wav
+m44,m44.wav,speaker4\speaker4_18.wav,speaker7\speaker7_03.wav
+m45,m45.wav,speaker8\speaker8_13.wav,speaker7\speaker7_15.wav
+m46,m46.wav,speaker1\speaker1_10.wav,speaker8\speaker8_19.wav
+m47,m47.wav,speaker7\speaker7_06.wav,speaker8\speaker8_14.wav
+m48,m48.wav,speaker10\speaker10_11.wav,speaker2\speaker2_09.wav
+m49,m49.wav,speaker3\speaker3_12.wav,speaker6\speaker6_08.wav
+m50,m50.wav,speaker1\speaker1_03.wav,speaker10\speaker10_08.wav

examples/manual_add_speaker.py

@@ -0,0 +1,62 @@
+"""手動語者辨識測試腳本
+
+此腳本允許一次針對某個 speaker 的多個檔案進行辨識，
+只要給定資料夾路徑和索引列表即可自動跑完。
+
+使用方式:
+    python -m examples.manual_add_speaker data/clean/speaker1 --indices 5 7 8
+    # 會自動處理 speaker1_05.wav, speaker1_07.wav, speaker1_08.wav
+"""
+
+import argparse
+from pathlib import Path
+from modules.identification import SpeakerIdentifier
+
+
+def main() -> None:
+    """解析參數並依序執行語者辨識。"""
+    parser = argparse.ArgumentParser(
+        description="針對某個 speaker 資料夾內指定編號的檔案，執行語者辨識"
+    )
+    parser.add_argument(
+        "speaker_dir",
+        help="speaker 資料夾路徑（如 data/clean/speaker1）"
+    )
+    parser.add_argument(
+        "--indices", "-i",
+        required=True,
+        nargs="+",
+        type=int,
+        metavar="N",
+        help="要處理的檔案編號列表（不含前綴零），如 5 7 8"
+    )
+    args = parser.parse_args()
+
+    speaker_path = Path(args.speaker_dir)
+    if not speaker_path.is_dir():
+        parser.error(f"{speaker_path!r} 不是一個有效的目錄")
+
+    speaker_name = speaker_path.name  # e.g. "speaker1"
+    identifier = SpeakerIdentifier()
+
+    for idx in args.indices:
+        # 兩位數補零
+        filename = f"{speaker_name}_{idx:02d}.wav"
+        audio_path = speaker_path / filename
+
+        print(f"\n🔍 處理：{audio_path}")
+        if not audio_path.is_file():
+            print(f"⚠️ 檔案不存在，跳過：{audio_path}")
+            continue
+
+        result = identifier.process_audio_file(str(audio_path))
+        if result:
+            speaker_id, speaker_label, distance = result
+            print(f"▶️ 語者: {speaker_label} (UUID: {speaker_id}) 相似度 {distance:.3f}")
+        else:
+            print("⚠️ 處理失敗或無法辨識")
+
+if __name__ == "__main__":
+    main()
+
+#python -m examples.manual_add_speaker data/clean/speaker1 -i 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20