{
  "quants": {
    "BF16": {
      "clipping_ratio": 0.0,
      "first_audio_seconds": null,
      "fixtures": 5,
      "frames_per_second": null,
      "intelligibility_metric": "WER for whitespace scripts; CER for CJK",
      "minutes_per_audio_hour": 264.0,
      "passed": true,
      "performance_from_review": true,
      "real_time_factor": 4.4,
      "results": [
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "first_output_seconds": null,
          "frames": 184320,
          "frames_per_second": null,
          "id": "numbers-punctuation-en",
          "minutes_per_audio_hour": null,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/BF16/00-numbers-punctuation-en.wav",
          "real_time_factor": null,
          "resumed_from_artifacts": true,
          "rms": 0.133046,
          "roundtrip_wer": 0.111111,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 3.84,
          "silence_ratio": 0.158523,
          "text": "Version 3.1 costs 42 dollars; is that correct?",
          "transcript": "Version 3.1 costs $42. Is that correct?"
        },
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "first_output_seconds": null,
          "frames": 384000,
          "frames_per_second": null,
          "id": "long-en",
          "minutes_per_audio_hour": null,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/BF16/01-long-en.wav",
          "real_time_factor": null,
          "resumed_from_artifacts": true,
          "rms": 0.204059,
          "roundtrip_wer": 0.0,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 8.0,
          "silence_ratio": 0.082812,
          "text": "A reliable voice release must produce clear audio, preserve punctuation and numbers, and include every companion file required by the runtime.",
          "transcript": "A reliable voice release must produce clear audio, preserve punctuation and numbers, and include every companion file required by the runtime."
        },
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "first_output_seconds": null,
          "frames": 145920,
          "frames_per_second": null,
          "id": "short-fr",
          "minutes_per_audio_hour": null,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/BF16/02-short-fr.wav",
          "real_time_factor": null,
          "resumed_from_artifacts": true,
          "rms": 0.160926,
          "roundtrip_wer": 0.142857,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 3.04,
          "silence_ratio": 0.097992,
          "text": "Bonjour, ceci est un test vocal multilingue.",
          "transcript": "Bonjour, ceci est un test vocal multilangue."
        },
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "first_output_seconds": null,
          "frames": 153600,
          "frames_per_second": null,
          "id": "short-ja",
          "minutes_per_audio_hour": null,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/BF16/03-short-ja.wav",
          "real_time_factor": null,
          "resumed_from_artifacts": true,
          "rms": 0.193536,
          "roundtrip_wer": 0.055556,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 3.2,
          "silence_ratio": 0.150664,
          "text": "\u3053\u3093\u306b\u3061\u306f\u3001\u3053\u308c\u306f\u97f3\u58f0\u5408\u6210\u306e\u30c6\u30b9\u30c8\u3067\u3059\u3002",
          "transcript": "\u3053\u3093\u306b\u3061\u306f\u3002\u3053\u308c\u306f\u97f3\u58f0\u5408\u58f0\u306e\u30c6\u30b9\u30c8\u3067\u3059\u3002"
        },
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "first_output_seconds": null,
          "frames": 192000,
          "frames_per_second": null,
          "id": "short-zh",
          "minutes_per_audio_hour": null,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/BF16/04-short-zh.wav",
          "real_time_factor": null,
          "resumed_from_artifacts": true,
          "rms": 0.099202,
          "roundtrip_wer": 0.5,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 4.0,
          "silence_ratio": 0.507698,
          "text": "\u4f60\u597d\uff0c\u8fd9\u662f\u4e00\u4e2a\u8bed\u97f3\u5408\u6210\u6d4b\u8bd5\u3002",
          "transcript": "\u4e5f\u597d\u9019\u662f\u4e00\u500b\u8a9e\u97f3\u5408\u6210\u6e2c\u8a66"
        }
      ],
      "resumed_from_artifacts": true,
      "roundtrip_wer": 0.161905,
      "silence_ratio": 0.507698
    },
    "F16": {
      "clipping_ratio": 0.0,
      "first_audio_seconds": 19.4324,
      "fixtures": 5,
      "frames_per_second": 10269.59,
      "intelligibility_metric": "WER for whitespace scripts; CER for CJK",
      "minutes_per_audio_hour": 290.35,
      "passed": true,
      "real_time_factor": 4.8391,
      "results": [
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "command": [
            "/home/linuxuser/.agentquantix/audio.cpp/build/bin/audiocpp_cli",
            "--task",
            "tts",
            "--family",
            "voxcpm2",
            "--model",
            "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/voxcpm2-F16.gguf",
            "--backend",
            "best",
            "--text",
            "Version 3.1 costs 42 dollars; is that correct?",
            "--out",
            "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/F16/00-numbers-punctuation-en.wav",
            "--metrics",
            "--language",
            "en"
          ],
          "elapsed_seconds": 18.3029,
          "first_output_seconds": 18.1222,
          "frames": 168960,
          "frames_per_second": 9231.32,
          "id": "numbers-punctuation-en",
          "minutes_per_audio_hour": 311.98,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/F16/00-numbers-punctuation-en.wav",
          "real_time_factor": 5.1997,
          "rms": 0.167574,
          "roundtrip_wer": 0.222222,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 3.52,
          "silence_ratio": 0.088394,
          "stdout": "family=voxcpm2\ntask=tts\nmode=offline\naudio_out=/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/F16/00-numbers-punctuation-en.wav\nmetrics.wall_ms=15655.9\nmetrics.audio_duration_ms=3520\nmetrics.rtf=4.4477\nmetrics.x_realtime=0.224835\nmetrics.audio_duration_source=output\nmetrics.sample_rate=48000\nmetrics.channels=1\n",
          "text": "Version 3.1 costs 42 dollars; is that correct?",
          "transcript": "Fursion 3.1 costs $42. Is that correct?"
        },
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "command": [
            "/home/linuxuser/.agentquantix/audio.cpp/build/bin/audiocpp_cli",
            "--task",
            "tts",
            "--family",
            "voxcpm2",
            "--model",
            "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/voxcpm2-F16.gguf",
            "--backend",
            "best",
            "--text",
            "A reliable voice release must produce clear audio, preserve punctuation and numbers, and include every companion file required by the runtime.",
            "--out",
            "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/F16/01-long-en.wav",
            "--metrics",
            "--language",
            "en"
          ],
          "elapsed_seconds": 26.7112,
          "first_output_seconds": 26.5305,
          "frames": 384000,
          "frames_per_second": 14375.99,
          "id": "long-en",
          "minutes_per_audio_hour": 200.33,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/F16/01-long-en.wav",
          "real_time_factor": 3.3389,
          "rms": 0.203471,
          "roundtrip_wer": 0.0,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 8.0,
          "silence_ratio": 0.083544,
          "stdout": "family=voxcpm2\ntask=tts\nmode=offline\naudio_out=/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/F16/01-long-en.wav\nmetrics.wall_ms=23994.4\nmetrics.audio_duration_ms=8000\nmetrics.rtf=2.99929\nmetrics.x_realtime=0.333412\nmetrics.audio_duration_source=output\nmetrics.sample_rate=48000\nmetrics.channels=1\n",
          "text": "A reliable voice release must produce clear audio, preserve punctuation and numbers, and include every companion file required by the runtime.",
          "transcript": "A reliable voice release must produce clear audio, preserve punctuation and numbers, and include every companion file required by the runtime."
        },
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "command": [
            "/home/linuxuser/.agentquantix/audio.cpp/build/bin/audiocpp_cli",
            "--task",
            "tts",
            "--family",
            "voxcpm2",
            "--model",
            "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/voxcpm2-F16.gguf",
            "--backend",
            "best",
            "--text",
            "Bonjour, ceci est un test vocal multilingue.",
            "--out",
            "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/F16/02-short-fr.wav",
            "--metrics",
            "--language",
            "fr"
          ],
          "elapsed_seconds": 17.4925,
          "first_output_seconds": 17.3117,
          "frames": 145920,
          "frames_per_second": 8341.86,
          "id": "short-fr",
          "minutes_per_audio_hour": 345.25,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/F16/02-short-fr.wav",
          "real_time_factor": 5.7541,
          "rms": 0.166584,
          "roundtrip_wer": 0.142857,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 3.04,
          "silence_ratio": 0.088172,
          "stdout": "family=voxcpm2\ntask=tts\nmode=offline\naudio_out=/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/F16/02-short-fr.wav\nmetrics.wall_ms=14803.8\nmetrics.audio_duration_ms=3040\nmetrics.rtf=4.86966\nmetrics.x_realtime=0.205353\nmetrics.audio_duration_source=output\nmetrics.sample_rate=48000\nmetrics.channels=1\n",
          "text": "Bonjour, ceci est un test vocal multilingue.",
          "transcript": "Bonjour ! Ceci est un test vocal multilangue."
        },
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "command": [
            "/home/linuxuser/.agentquantix/audio.cpp/build/bin/audiocpp_cli",
            "--task",
            "tts",
            "--family",
            "voxcpm2",
            "--model",
            "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/voxcpm2-F16.gguf",
            "--backend",
            "best",
            "--text",
            "\u3053\u3093\u306b\u3061\u306f\u3001\u3053\u308c\u306f\u97f3\u58f0\u5408\u6210\u306e\u30c6\u30b9\u30c8\u3067\u3059\u3002",
            "--out",
            "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/F16/03-short-ja.wav",
            "--metrics",
            "--language",
            "ja"
          ],
          "elapsed_seconds": 16.2086,
          "first_output_seconds": 16.0279,
          "frames": 153600,
          "frames_per_second": 9476.45,
          "id": "short-ja",
          "minutes_per_audio_hour": 303.91,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/F16/03-short-ja.wav",
          "real_time_factor": 5.0652,
          "rms": 0.194407,
          "roundtrip_wer": 0.055556,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 3.2,
          "silence_ratio": 0.162826,
          "stdout": "family=voxcpm2\ntask=tts\nmode=offline\naudio_out=/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/F16/03-short-ja.wav\nmetrics.wall_ms=13493.4\nmetrics.audio_duration_ms=3200\nmetrics.rtf=4.21668\nmetrics.x_realtime=0.237153\nmetrics.audio_duration_source=output\nmetrics.sample_rate=48000\nmetrics.channels=1\n",
          "text": "\u3053\u3093\u306b\u3061\u306f\u3001\u3053\u308c\u306f\u97f3\u58f0\u5408\u6210\u306e\u30c6\u30b9\u30c8\u3067\u3059\u3002",
          "transcript": "\u3053\u3093\u306b\u3061\u306f\u3002\u3053\u308c\u306f\u97f3\u58f0\u5408\u6027\u306e\u30c6\u30b9\u30c8\u3067\u3059\u3002"
        },
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "command": [
            "/home/linuxuser/.agentquantix/audio.cpp/build/bin/audiocpp_cli",
            "--task",
            "tts",
            "--family",
            "voxcpm2",
            "--model",
            "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/voxcpm2-F16.gguf",
            "--backend",
            "best",
            "--text",
            "\u4f60\u597d\uff0c\u8fd9\u662f\u4e00\u4e2a\u8bed\u97f3\u5408\u6210\u6d4b\u8bd5\u3002",
            "--out",
            "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/F16/04-short-zh.wav",
            "--metrics",
            "--language",
            "zh"
          ],
          "elapsed_seconds": 19.3503,
          "first_output_seconds": 19.1696,
          "frames": 192000,
          "frames_per_second": 9922.33,
          "id": "short-zh",
          "minutes_per_audio_hour": 290.26,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/F16/04-short-zh.wav",
          "real_time_factor": 4.8376,
          "rms": 0.097418,
          "roundtrip_wer": 0.5,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 4.0,
          "silence_ratio": 0.506927,
          "stdout": "family=voxcpm2\ntask=tts\nmode=offline\naudio_out=/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/F16/04-short-zh.wav\nmetrics.wall_ms=16673.3\nmetrics.audio_duration_ms=4000\nmetrics.rtf=4.16832\nmetrics.x_realtime=0.239905\nmetrics.audio_duration_source=output\nmetrics.sample_rate=48000\nmetrics.channels=1\n",
          "text": "\u4f60\u597d\uff0c\u8fd9\u662f\u4e00\u4e2a\u8bed\u97f3\u5408\u6210\u6d4b\u8bd5\u3002",
          "transcript": "\u4e5f\u597d\u9019\u662f\u4e00\u500b\u8a9e\u97f3\u5408\u6210\u6e2c\u8a66"
        }
      ],
      "roundtrip_wer": 0.184127,
      "silence_ratio": 0.506927
    },
    "Q3_K": {
      "clipping_ratio": 0.0,
      "first_audio_seconds": null,
      "fixtures": 5,
      "frames_per_second": null,
      "intelligibility_metric": "WER for whitespace scripts; CER for CJK",
      "minutes_per_audio_hour": 324.0,
      "passed": true,
      "performance_from_review": true,
      "real_time_factor": 5.4,
      "results": [
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "first_output_seconds": null,
          "frames": 192000,
          "frames_per_second": null,
          "id": "numbers-punctuation-en",
          "minutes_per_audio_hour": null,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/Q3_K/00-numbers-punctuation-en.wav",
          "real_time_factor": null,
          "resumed_from_artifacts": true,
          "rms": 0.326681,
          "roundtrip_wer": 0.111111,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 4.0,
          "silence_ratio": 0.025245,
          "text": "Version 3.1 costs 42 dollars; is that correct?",
          "transcript": "version 3.1 cost 42 dollars. Is that correct?"
        },
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "first_output_seconds": null,
          "frames": 453120,
          "frames_per_second": null,
          "id": "long-en",
          "minutes_per_audio_hour": null,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/Q3_K/01-long-en.wav",
          "real_time_factor": null,
          "resumed_from_artifacts": true,
          "rms": 0.380062,
          "roundtrip_wer": 0.047619,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 9.44,
          "silence_ratio": 0.034194,
          "text": "A reliable voice release must produce clear audio, preserve punctuation and numbers, and include every companion file required by the runtime.",
          "transcript": "A reliable voice release must produce clear audio, preserve punctuation and numbers, and include every companion files required by the runtime."
        },
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "first_output_seconds": null,
          "frames": 192000,
          "frames_per_second": null,
          "id": "short-fr",
          "minutes_per_audio_hour": null,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/Q3_K/02-short-fr.wav",
          "real_time_factor": null,
          "resumed_from_artifacts": true,
          "rms": 0.272483,
          "roundtrip_wer": 0.428571,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 4.0,
          "silence_ratio": 0.039297,
          "text": "Bonjour, ceci est un test vocal multilingue.",
          "transcript": "Bonjour ! Ceci est un des vacales m\u00e9decines."
        },
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "first_output_seconds": null,
          "frames": 153600,
          "frames_per_second": null,
          "id": "short-ja",
          "minutes_per_audio_hour": null,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/Q3_K/03-short-ja.wav",
          "real_time_factor": null,
          "resumed_from_artifacts": true,
          "rms": 0.307624,
          "roundtrip_wer": 0.055556,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 3.2,
          "silence_ratio": 0.126797,
          "text": "\u3053\u3093\u306b\u3061\u306f\u3001\u3053\u308c\u306f\u97f3\u58f0\u5408\u6210\u306e\u30c6\u30b9\u30c8\u3067\u3059\u3002",
          "transcript": "\u3053\u3093\u306b\u3061\u306f\u3002\u3053\u308c\u306f\u97f3\u58f0\u5408\u58f0\u306e\u30c6\u30b9\u30c8\u3067\u3059\u3002"
        },
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "first_output_seconds": null,
          "frames": 153600,
          "frames_per_second": null,
          "id": "short-zh",
          "minutes_per_audio_hour": null,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/Q3_K/04-short-zh.wav",
          "real_time_factor": null,
          "resumed_from_artifacts": true,
          "rms": 0.186435,
          "roundtrip_wer": 0.333333,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 3.2,
          "silence_ratio": 0.233431,
          "text": "\u4f60\u597d\uff0c\u8fd9\u662f\u4e00\u4e2a\u8bed\u97f3\u5408\u6210\u6d4b\u8bd5\u3002",
          "transcript": "\u4f60\u597d,\u8fd9\u662f\u4e00\u4e2a\u9c7c\u548c\u57ce\u6d4b\u8bd5"
        }
      ],
      "resumed_from_artifacts": true,
      "roundtrip_wer": 0.195238,
      "silence_ratio": 0.233431
    },
    "Q4_K": {
      "clipping_ratio": 0.0,
      "first_audio_seconds": null,
      "fixtures": 5,
      "frames_per_second": null,
      "intelligibility_metric": "WER for whitespace scripts; CER for CJK",
      "minutes_per_audio_hour": 300.0,
      "passed": true,
      "performance_from_review": true,
      "real_time_factor": 5.0,
      "results": [
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "first_output_seconds": null,
          "frames": 161280,
          "frames_per_second": null,
          "id": "numbers-punctuation-en",
          "minutes_per_audio_hour": null,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/Q4_K/00-numbers-punctuation-en.wav",
          "real_time_factor": null,
          "resumed_from_artifacts": true,
          "rms": 0.198127,
          "roundtrip_wer": 0.111111,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 3.36,
          "silence_ratio": 0.084524,
          "text": "Version 3.1 costs 42 dollars; is that correct?",
          "transcript": "Version 3.1 costs $42. Is that correct?"
        },
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "first_output_seconds": null,
          "frames": 376320,
          "frames_per_second": null,
          "id": "long-en",
          "minutes_per_audio_hour": null,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/Q4_K/01-long-en.wav",
          "real_time_factor": null,
          "resumed_from_artifacts": true,
          "rms": 0.196146,
          "roundtrip_wer": 0.0,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 7.84,
          "silence_ratio": 0.07702,
          "text": "A reliable voice release must produce clear audio, preserve punctuation and numbers, and include every companion file required by the runtime.",
          "transcript": "A reliable voice release must produce clear audio, preserve punctuation and numbers, and include every companion file required by the runtime."
        },
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "first_output_seconds": null,
          "frames": 168960,
          "frames_per_second": null,
          "id": "short-fr",
          "minutes_per_audio_hour": null,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/Q4_K/02-short-fr.wav",
          "real_time_factor": null,
          "resumed_from_artifacts": true,
          "rms": 0.131053,
          "roundtrip_wer": 0.142857,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 3.52,
          "silence_ratio": 0.075059,
          "text": "Bonjour, ceci est un test vocal multilingue.",
          "transcript": "Bonjour ! Ceci est un test vocal multilangue !"
        },
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "first_output_seconds": null,
          "frames": 130560,
          "frames_per_second": null,
          "id": "short-ja",
          "minutes_per_audio_hour": null,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/Q4_K/03-short-ja.wav",
          "real_time_factor": null,
          "resumed_from_artifacts": true,
          "rms": 0.23994,
          "roundtrip_wer": 0.055556,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 2.72,
          "silence_ratio": 0.128661,
          "text": "\u3053\u3093\u306b\u3061\u306f\u3001\u3053\u308c\u306f\u97f3\u58f0\u5408\u6210\u306e\u30c6\u30b9\u30c8\u3067\u3059\u3002",
          "transcript": "\u3053\u3093\u306b\u3061\u306f\u3002\u3053\u308c\u306f\u97f3\u58f0\u5408\u58f0\u306e\u30c6\u30b9\u30c8\u3067\u3059\u3002"
        },
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "first_output_seconds": null,
          "frames": 153600,
          "frames_per_second": null,
          "id": "short-zh",
          "minutes_per_audio_hour": null,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/Q4_K/04-short-zh.wav",
          "real_time_factor": null,
          "resumed_from_artifacts": true,
          "rms": 0.163529,
          "roundtrip_wer": 0.25,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 3.2,
          "silence_ratio": 0.244212,
          "text": "\u4f60\u597d\uff0c\u8fd9\u662f\u4e00\u4e2a\u8bed\u97f3\u5408\u6210\u6d4b\u8bd5\u3002",
          "transcript": "\u60a8\u597d \u8fd9\u662f\u4e00\u4e2a\u4e0e\u4e91\u5408\u6210\u6d4b\u8bd5"
        }
      ],
      "resumed_from_artifacts": true,
      "roundtrip_wer": 0.111905,
      "silence_ratio": 0.244212
    },
    "Q5_K": {
      "clipping_ratio": 0.0,
      "first_audio_seconds": null,
      "fixtures": 5,
      "frames_per_second": null,
      "intelligibility_metric": "WER for whitespace scripts; CER for CJK",
      "minutes_per_audio_hour": 342.0,
      "passed": true,
      "performance_from_review": true,
      "real_time_factor": 5.7,
      "results": [
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "first_output_seconds": null,
          "frames": 238080,
          "frames_per_second": null,
          "id": "numbers-punctuation-en",
          "minutes_per_audio_hour": null,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/Q5_K/00-numbers-punctuation-en.wav",
          "real_time_factor": null,
          "resumed_from_artifacts": true,
          "rms": 0.120141,
          "roundtrip_wer": 0.111111,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 4.96,
          "silence_ratio": 0.153385,
          "text": "Version 3.1 costs 42 dollars; is that correct?",
          "transcript": "version 3.1 costs $42. Is that correct?"
        },
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "first_output_seconds": null,
          "frames": 384000,
          "frames_per_second": null,
          "id": "long-en",
          "minutes_per_audio_hour": null,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/Q5_K/01-long-en.wav",
          "real_time_factor": null,
          "resumed_from_artifacts": true,
          "rms": 0.19138,
          "roundtrip_wer": 0.0,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 8.0,
          "silence_ratio": 0.089805,
          "text": "A reliable voice release must produce clear audio, preserve punctuation and numbers, and include every companion file required by the runtime.",
          "transcript": "A reliable voice release must produce clear audio, preserve punctuation and numbers, and include every companion file required by the runtime."
        },
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "first_output_seconds": null,
          "frames": 192000,
          "frames_per_second": null,
          "id": "short-fr",
          "minutes_per_audio_hour": null,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/Q5_K/02-short-fr.wav",
          "real_time_factor": null,
          "resumed_from_artifacts": true,
          "rms": 0.163489,
          "roundtrip_wer": 0.142857,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 4.0,
          "silence_ratio": 0.110068,
          "text": "Bonjour, ceci est un test vocal multilingue.",
          "transcript": "Bonjour ! Ceci est un test vocal multilangue."
        },
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "first_output_seconds": null,
          "frames": 130560,
          "frames_per_second": null,
          "id": "short-ja",
          "minutes_per_audio_hour": null,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/Q5_K/03-short-ja.wav",
          "real_time_factor": null,
          "resumed_from_artifacts": true,
          "rms": 0.174939,
          "roundtrip_wer": 0.111111,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 2.72,
          "silence_ratio": 0.088496,
          "text": "\u3053\u3093\u306b\u3061\u306f\u3001\u3053\u308c\u306f\u97f3\u58f0\u5408\u6210\u306e\u30c6\u30b9\u30c8\u3067\u3059\u3002",
          "transcript": "\u3053\u3093\u306b\u3061\u306f\u3002\u3053\u308c\u306f\u97f3\u58f0\u9ad8\u58f0\u306e\u30c6\u30b9\u30c8\u3067\u3059\u3002"
        },
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "first_output_seconds": null,
          "frames": 161280,
          "frames_per_second": null,
          "id": "short-zh",
          "minutes_per_audio_hour": null,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/Q5_K/04-short-zh.wav",
          "real_time_factor": null,
          "resumed_from_artifacts": true,
          "rms": 0.101781,
          "roundtrip_wer": 0.166667,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 3.36,
          "silence_ratio": 0.456262,
          "text": "\u4f60\u597d\uff0c\u8fd9\u662f\u4e00\u4e2a\u8bed\u97f3\u5408\u6210\u6d4b\u8bd5\u3002",
          "transcript": "\u4f60\u597d \u8fd9\u662f\u4e00\u4e2a\u8bed\u97f3\u548c\u8bda\u6d4b\u8bd5"
        }
      ],
      "resumed_from_artifacts": true,
      "roundtrip_wer": 0.106349,
      "silence_ratio": 0.456262
    },
    "Q6_K": {
      "clipping_ratio": 0.0,
      "first_audio_seconds": null,
      "fixtures": 5,
      "frames_per_second": null,
      "intelligibility_metric": "WER for whitespace scripts; CER for CJK",
      "minutes_per_audio_hour": 300.0,
      "passed": true,
      "performance_from_review": true,
      "real_time_factor": 5.0,
      "results": [
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "first_output_seconds": null,
          "frames": 207360,
          "frames_per_second": null,
          "id": "numbers-punctuation-en",
          "minutes_per_audio_hour": null,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/Q6_K/00-numbers-punctuation-en.wav",
          "real_time_factor": null,
          "resumed_from_artifacts": true,
          "rms": 0.157859,
          "roundtrip_wer": 0.111111,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 4.32,
          "silence_ratio": 0.193475,
          "text": "Version 3.1 costs 42 dollars; is that correct?",
          "transcript": "Version 3.1 costs $42. Is that correct?"
        },
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "first_output_seconds": null,
          "frames": 384000,
          "frames_per_second": null,
          "id": "long-en",
          "minutes_per_audio_hour": null,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/Q6_K/01-long-en.wav",
          "real_time_factor": null,
          "resumed_from_artifacts": true,
          "rms": 0.21542,
          "roundtrip_wer": 0.0,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 8.0,
          "silence_ratio": 0.073685,
          "text": "A reliable voice release must produce clear audio, preserve punctuation and numbers, and include every companion file required by the runtime.",
          "transcript": "A reliable voice release must produce clear audio, preserve punctuation and numbers, and include every companion file required by the runtime."
        },
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "first_output_seconds": null,
          "frames": 153600,
          "frames_per_second": null,
          "id": "short-fr",
          "minutes_per_audio_hour": null,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/Q6_K/02-short-fr.wav",
          "real_time_factor": null,
          "resumed_from_artifacts": true,
          "rms": 0.188378,
          "roundtrip_wer": 0.142857,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 3.2,
          "silence_ratio": 0.055319,
          "text": "Bonjour, ceci est un test vocal multilingue.",
          "transcript": "Bonjour ! Ceci est un test vocal multilangue !"
        },
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "first_output_seconds": null,
          "frames": 161280,
          "frames_per_second": null,
          "id": "short-ja",
          "minutes_per_audio_hour": null,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/Q6_K/03-short-ja.wav",
          "real_time_factor": null,
          "resumed_from_artifacts": true,
          "rms": 0.198035,
          "roundtrip_wer": 0.055556,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 3.36,
          "silence_ratio": 0.124709,
          "text": "\u3053\u3093\u306b\u3061\u306f\u3001\u3053\u308c\u306f\u97f3\u58f0\u5408\u6210\u306e\u30c6\u30b9\u30c8\u3067\u3059\u3002",
          "transcript": "\u3053\u3093\u306b\u3061\u306f\u3002\u3053\u308c\u306f\u97f3\u58f0\u5408\u6027\u306e\u30c6\u30b9\u30c8\u3067\u3059\u3002"
        },
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "first_output_seconds": null,
          "frames": 215040,
          "frames_per_second": null,
          "id": "short-zh",
          "minutes_per_audio_hour": null,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/Q6_K/04-short-zh.wav",
          "real_time_factor": null,
          "resumed_from_artifacts": true,
          "rms": 0.128449,
          "roundtrip_wer": 0.5,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 4.48,
          "silence_ratio": 0.510324,
          "text": "\u4f60\u597d\uff0c\u8fd9\u662f\u4e00\u4e2a\u8bed\u97f3\u5408\u6210\u6d4b\u8bd5\u3002",
          "transcript": "\u4e5f\u597d \u9019\u662f\u4e00\u500b\u8a9e\u97f3\u5408\u6210\u6e2c\u8a66"
        }
      ],
      "resumed_from_artifacts": true,
      "roundtrip_wer": 0.161905,
      "silence_ratio": 0.510324
    },
    "Q8_0": {
      "clipping_ratio": 0.0,
      "first_audio_seconds": null,
      "fixtures": 5,
      "frames_per_second": null,
      "intelligibility_metric": "WER for whitespace scripts; CER for CJK",
      "minutes_per_audio_hour": 300.0,
      "passed": true,
      "performance_from_review": true,
      "real_time_factor": 5.0,
      "results": [
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "first_output_seconds": null,
          "frames": 184320,
          "frames_per_second": null,
          "id": "numbers-punctuation-en",
          "minutes_per_audio_hour": null,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/Q8_0/00-numbers-punctuation-en.wav",
          "real_time_factor": null,
          "resumed_from_artifacts": true,
          "rms": 0.179811,
          "roundtrip_wer": 0.111111,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 3.84,
          "silence_ratio": 0.148492,
          "text": "Version 3.1 costs 42 dollars; is that correct?",
          "transcript": "version 3.1 costs $42. Is that correct?"
        },
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "first_output_seconds": null,
          "frames": 384000,
          "frames_per_second": null,
          "id": "long-en",
          "minutes_per_audio_hour": null,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/Q8_0/01-long-en.wav",
          "real_time_factor": null,
          "resumed_from_artifacts": true,
          "rms": 0.20168,
          "roundtrip_wer": 0.0,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 8.0,
          "silence_ratio": 0.07607,
          "text": "A reliable voice release must produce clear audio, preserve punctuation and numbers, and include every companion file required by the runtime.",
          "transcript": "A reliable voice release must produce clear audio, preserve punctuation and numbers, and include every companion file required by the runtime."
        },
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "first_output_seconds": null,
          "frames": 145920,
          "frames_per_second": null,
          "id": "short-fr",
          "minutes_per_audio_hour": null,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/Q8_0/02-short-fr.wav",
          "real_time_factor": null,
          "resumed_from_artifacts": true,
          "rms": 0.179413,
          "roundtrip_wer": 0.142857,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 3.04,
          "silence_ratio": 0.067989,
          "text": "Bonjour, ceci est un test vocal multilingue.",
          "transcript": "Bonjour ! Ceci est un test vocal multilangue."
        },
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "first_output_seconds": null,
          "frames": 153600,
          "frames_per_second": null,
          "id": "short-ja",
          "minutes_per_audio_hour": null,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/Q8_0/03-short-ja.wav",
          "real_time_factor": null,
          "resumed_from_artifacts": true,
          "rms": 0.202345,
          "roundtrip_wer": 0.055556,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 3.2,
          "silence_ratio": 0.166836,
          "text": "\u3053\u3093\u306b\u3061\u306f\u3001\u3053\u308c\u306f\u97f3\u58f0\u5408\u6210\u306e\u30c6\u30b9\u30c8\u3067\u3059\u3002",
          "transcript": "\u3053\u3093\u306b\u3061\u306f\u3002\u3053\u308c\u306f\u97f3\u58f0\u5408\u6027\u306e\u30c6\u30b9\u30c8\u3067\u3059\u3002"
        },
        {
          "channels": 1,
          "clipping_ratio": 0.0,
          "first_output_seconds": null,
          "frames": 215040,
          "frames_per_second": null,
          "id": "short-zh",
          "minutes_per_audio_hour": null,
          "path": "/home/linuxuser/.agentquantix/temp/voice/openbmb--VoxCPM2/models/quality/Q8_0/04-short-zh.wav",
          "real_time_factor": null,
          "resumed_from_artifacts": true,
          "rms": 0.120478,
          "roundtrip_wer": 0.5,
          "sample_rate": 48000,
          "sample_width": 2,
          "seconds": 4.48,
          "silence_ratio": 0.497661,
          "text": "\u4f60\u597d\uff0c\u8fd9\u662f\u4e00\u4e2a\u8bed\u97f3\u5408\u6210\u6d4b\u8bd5\u3002",
          "transcript": "\u4e5f\u597d\u9019\u662f\u4e00\u500b\u8a9e\u97f3\u5408\u6210\u6e2c\u8a66"
        }
      ],
      "resumed_from_artifacts": true,
      "roundtrip_wer": 0.161905,
      "silence_ratio": 0.497661
    }
  }
}