Evaluation Report: HK 4-array v3.2 vs CAE v1.2 Microphones
Report date: 2026-06-09
1. Objective
This report compares two robot microphone options: HK mic 4 array linear v3.2 and CAE v1.2. Each dataset contains the real test microphone and an obs laptop-microphone recording captured as a simultaneous reference.
The test conditions cover 1 m, 2 m, and 3 m speech distance. For each distance, recordings were made without robot-speaker playback and with robot-speaker playback to evaluate echo robustness.
2. Input Data
- HK real microphone:
/home/thachhs/end_baocaomic/michk_32/hk/<1m|2m|3m>/*.wav. - HK laptop reference:
/home/thachhs/end_baocaomic/michk_32/obs/<1m|2m|3m>/*.wav. - CAE real microphone:
/home/thachhs/end_baocaomic/micanhthao/micmoi/<1m|2m|3m>/*.wav. - CAE laptop reference:
/home/thachhs/end_baocaomic/micanhthao/obs/<1m|2m|3m>/*.wav.
Files ending in _echo are treated as robot-speaker playback / echo cases. For example, 1m/1_echo.wav from the real microphone is paired with 1m/1_echo.wav under the matching obs folder.
3. Technical Metrics
- LUFS: integrated loudness, used to compare perceived recording level.
- RMS/Peak dBFS: average and peak level relative to full scale.
- Estimated SNR: active RMS minus noise floor; the noise floor is estimated from the lowest-energy 20% of frames.
- Absolute active %: percentage of frames above -50 dBFS; useful for detecting suppressed or missing audio.
- Leading silence ms: initial near-silent duration below -55 dBFS.
- Longest silence ms: longest near-silent interval below -55 dBFS.
- Dropout score: 0-100 score combining silence ratio, long gaps, start loss, and near-muted files.
- Echo risk flags: echo files are flagged if loudness drops by at least 5 LU against the same-distance no-echo baseline, active ratio is below 45%, speech-band energy is below 30%, SNR is below 28 dB, envelope correlation with OBS is below 0.30, duration is abnormal, or an energy dropout is detected.
- Speech %: energy share in the 300-3400 Hz speech band.
- Echo index: envelope autocorrelation heuristic in the 50-500 ms lag region; used only for relative comparison because no separate far-end signal is available for true ERLE.
- LSD dB: log-spectral distance between the real microphone and laptop reference within the same paired recording.
4. Executive Summary
- Without robot speaker playback, both microphones captured usable speech: average SNR is 54.8 dB for HK and 39.1 dB for CAE.
- With robot-speaker echo, HK is more stable on the key indicators: HK echo SNR is 42.0 dB versus 31.5 dB for CAE; HK echo active ratio is 71.8% versus 65.6% for CAE; HK/OBS envelope correlation is 0.506 versus 0.421 for CAE/OBS.
- CAE shows stronger suppression / information-loss risk in echo: CAE echo LUFS drops by -4.4 LU versus its no-echo baseline at the same distance, while HK changes by +4.0 LU.
- Echo files with at least two risk flags: HK 0/15, CAE 8/21. Low-loudness echo files: HK 0, CAE 7; low-active files: HK 0, CAE 4; abnormally low speech-band files: HK 0, CAE 1.
- Average echo loudness is -23.6 LUFS for HK and -24.7 LUFS for CAE. No meaningful clipping was detected on either microphone.
- Deployment favors HK: it is compact and includes tuning software; CAE requires an external sound card and a more complex setup flow.
5. Overall HK vs CAE Summary
| Mic | N normal | N echo | Normal SNR dB | Echo SNR dB | Normal LUFS | Echo LUFS | Echo dropout score | Echo start-loss files | Echo long-dropout files | Echo muted/low files | Clip max % | Format | Deployment note |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| HK 4-array v3.2 | 15 | 15 | 54.8 ± 7.1 | 42.0 ± 11.0 | -27.6 ± 1.2 | -23.6 ± 2.1 | 11.6 ± 6.6 | 2 | 3 | 0 | 0 | 16000 Hz, 2 ch (30) | Compact, with built-in tuning/configuration software. |
| CAE v1.2 | 30 | 21 | 39.1 ± 1.3 | 31.5 ± 3.8 | -20.3 ± 0.8 | -24.7 ± 3.0 | 9.1 ± 5.6 | 1 | 1 | 0 | 0 | 16000 Hz, 1 ch (51) | More complex installation, requiring an external sound card. |
6. Real-Microphone Comparison by Distance and Mode
| Mode | Distance | Mic | N | LUFS | RMS dBFS | SNR dB | Abs. active % | Leading silence ms | Longest silence ms | Dropout score | Muted/low files | Speech % | Echo index | Clip % |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Normal | 1m | HK 4-array v3.2 | 5 | -27.0 ± 0.6 | -27.8 ± 1.4 | 52.1 ± 10.5 | 65.9 ± 12.9 | 512 ± 590 | 1006 ± 366 | 21.2 ± 10.2 | 0 | 67.9 ± 10.9 | 0.868 ± 0.054 | 0.000 ± 0.000 |
| Normal | 1m | CAE v1.2 | 10 | -20.1 ± 0.4 | -21.3 ± 0.8 | 39.3 ± 1.4 | 60.4 ± 11.3 | 326 ± 270 | 716 ± 294 | 20.1 ± 6.6 | 0 | 52.5 ± 8.2 | 0.898 ± 0.028 | 0.000 ± 0.000 |
| Normal | 2m | HK 4-array v3.2 | 5 | -28.2 ± 1.8 | -28.8 ± 1.9 | 54.7 ± 6.5 | 65.0 ± 8.5 | 104 ± 108 | 1024 ± 81 | 20.6 ± 5.0 | 0 | 80.5 ± 6.2 | 0.853 ± 0.028 | 0.000 ± 0.000 |
| Normal | 2m | CAE v1.2 | 10 | -20.4 ± 1.1 | -21.6 ± 1.2 | 38.8 ± 1.7 | 63.5 ± 9.6 | 230 ± 184 | 698 ± 346 | 17.8 ± 5.9 | 0 | 60.0 ± 9.1 | 0.846 ± 0.059 | 0.000 ± 0.000 |
| Normal | 3m | HK 4-array v3.2 | 5 | -27.6 ± 0.8 | -28.2 ± 0.9 | 57.5 ± 1.9 | 69.4 ± 3.9 | 256 ± 266 | 886 ± 230 | 18.1 ± 2.1 | 0 | 91.2 ± 3.1 | 0.864 ± 0.039 | 0.000 ± 0.000 |
| Normal | 3m | CAE v1.2 | 10 | -20.3 ± 1.0 | -21.9 ± 1.2 | 39.2 ± 0.9 | 58.4 ± 5.7 | 237 ± 234 | 962 ± 428 | 20.5 ± 5.4 | 0 | 51.0 ± 5.3 | 0.897 ± 0.022 | 0.000 ± 0.000 |
| Echo | 1m | HK 4-array v3.2 | 5 | -21.5 ± 1.2 | -23.0 ± 1.2 | 41.1 ± 10.9 | 73.1 ± 7.6 | 32 ± 72 | 296 ± 214 | 9.9 ± 4.8 | 0 | 84.5 ± 3.7 | 0.632 ± 0.103 | 0.000 ± 0.000 |
| Echo | 1m | CAE v1.2 | 11 | -25.0 ± 3.6 | -27.0 ± 3.7 | 31.9 ± 4.9 | 60.1 ± 22.3 | 54 ± 94 | 399 ± 353 | 11.0 ± 6.6 | 0 | 85.5 ± 26.4 | 0.766 ± 0.140 | 0.000 ± 0.000 |
| Echo | 2m | HK 4-array v3.2 | 5 | -24.4 ± 2.2 | -25.1 ± 1.8 | 37.7 ± 12.2 | 76.1 ± 9.7 | 112 ± 155 | 396 ± 470 | 9.8 ± 7.7 | 0 | 88.2 ± 6.3 | 0.619 ± 0.161 | 0.000 ± 0.000 |
| Echo | 2m | CAE v1.2 | 5 | -24.9 ± 2.6 | -26.3 ± 3.0 | 29.3 ± 1.9 | 78.0 ± 11.9 | 0 ± 0 | 152 ± 153 | 4.4 ± 3.0 | 0 | 89.8 ± 3.9 | 0.768 ± 0.064 | 0.000 ± 0.000 |
| Echo | 3m | HK 4-array v3.2 | 5 | -24.9 ± 1.1 | -26.5 ± 1.2 | 47.3 ± 10.0 | 66.3 ± 6.5 | 228 ± 482 | 576 ± 419 | 15.2 ± 6.9 | 0 | 91.2 ± 3.1 | 0.721 ± 0.125 | 0.000 ± 0.000 |
| Echo | 3m | CAE v1.2 | 5 | -23.8 ± 1.9 | -26.6 ± 2.0 | 32.7 ± 1.2 | 65.3 ± 6.6 | 64 ± 87 | 284 ± 31 | 9.4 ± 2.0 | 0 | 97.1 ± 1.0 | 0.772 ± 0.040 | 0.000 ± 0.000 |
7. Real Microphone vs Laptop/OBS Paired Comparison
Positive Mic-OBS values mean the real microphone is higher than the laptop microphone. Higher Envelope corr means the two sources capture the same acoustic events more similarly after time alignment.
| Mode | Distance | Mic | N pairs | Mic-OBS RMS dB | Mic-OBS SNR dB | Active delta pp | Dropout delta | Envelope corr | LSD dB |
|---|---|---|---|---|---|---|---|---|---|
| Normal | 1m | CAE v1.2 | 10 | 6.4 ± 0.8 | 27.1 ± 1.3 | -39.6 ± 11.3 | 20.1 ± 6.6 | 0.566 ± 0.132 | 6.7 ± 0.7 |
| Normal | 1m | HK 4-array v3.2 | 5 | -2.9 ± 0.7 | 40.6 ± 10.5 | -34.1 ± 12.9 | 21.2 ± 10.2 | 0.670 ± 0.221 | 8.5 ± 0.5 |
| Normal | 2m | CAE v1.2 | 10 | 5.6 ± 2.6 | 26.2 ± 2.3 | -36.5 ± 9.6 | 17.8 ± 5.9 | 0.598 ± 0.161 | 6.7 ± 0.4 |
| Normal | 2m | HK 4-array v3.2 | 5 | -4.2 ± 1.3 | 43.0 ± 6.4 | -35.0 ± 8.5 | 20.6 ± 5.0 | 0.769 ± 0.052 | 10.3 ± 0.9 |
| Normal | 3m | CAE v1.2 | 10 | 6.2 ± 1.0 | 27.7 ± 1.4 | -41.6 ± 5.7 | 20.5 ± 5.4 | 0.652 ± 0.130 | 6.3 ± 0.6 |
| Normal | 3m | HK 4-array v3.2 | 5 | -4.0 ± 0.9 | 45.9 ± 1.9 | -30.6 ± 3.9 | 18.1 ± 2.1 | 0.775 ± 0.026 | 9.2 ± 0.7 |
| Echo | 1m | CAE v1.2 | 11 | -17.5 ± 3.9 | 11.9 ± 6.8 | -39.9 ± 22.3 | 11.0 ± 6.6 | 0.422 ± 0.193 | 8.9 ± 1.9 |
| Echo | 1m | HK 4-array v3.2 | 5 | -15.6 ± 1.3 | 22.0 ± 9.4 | -26.9 ± 7.6 | 9.9 ± 4.8 | 0.479 ± 0.092 | 11.1 ± 0.9 |
| Echo | 2m | CAE v1.2 | 5 | -17.3 ± 3.0 | 13.3 ± 2.8 | -22.0 ± 11.9 | 4.4 ± 3.0 | 0.348 ± 0.082 | 7.7 ± 1.3 |
| Echo | 2m | HK 4-array v3.2 | 5 | -18.0 ± 2.0 | 17.5 ± 9.2 | -23.9 ± 9.7 | 9.8 ± 7.7 | 0.516 ± 0.182 | 8.3 ± 1.4 |
| Echo | 3m | CAE v1.2 | 5 | -17.4 ± 2.3 | 15.7 ± 3.1 | -34.7 ± 6.6 | 9.4 ± 2.0 | 0.492 ± 0.177 | 10.3 ± 2.5 |
| Echo | 3m | HK 4-array v3.2 | 5 | -19.1 ± 1.4 | 27.1 ± 10.0 | -33.7 ± 6.5 | 15.2 ± 6.9 | 0.524 ± 0.206 | 8.2 ± 0.8 |
8. Echo Impact vs No-Echo Condition
Delta columns are computed as the echo-condition mean minus the no-echo-condition mean for the same microphone and distance.
| Mic | Distance | N normal | N echo | Delta LUFS | Delta RMS dB | Delta SNR dB | Delta active pp | Delta dropout | Delta leading ms | Delta echo index |
|---|---|---|---|---|---|---|---|---|---|---|
| HK 4-array v3.2 | 1m | 5 | 5 | 5.417 | 4.799 | -11.067 | 7.263 | -11.326 | -480 | -0.236 |
| HK 4-array v3.2 | 2m | 5 | 5 | 3.866 | 3.721 | -17.009 | 11.113 | -10.829 | 8 | -0.234 |
| HK 4-array v3.2 | 3m | 5 | 5 | 2.699 | 1.708 | -10.201 | -3.084 | -2.943 | -28 | -0.143 |
| CAE v1.2 | 1m | 10 | 11 | -4.846 | -5.71 | -7.414 | -0.24 | -9.086 | -272.364 | -0.132 |
| CAE v1.2 | 2m | 10 | 5 | -4.472 | -4.734 | -9.558 | 14.549 | -13.427 | -230 | -0.078 |
| CAE v1.2 | 3m | 10 | 5 | -3.516 | -4.707 | -6.548 | 6.961 | -11.053 | -173 | -0.125 |
9. Echo Information-Loss Risk
This section focuses on indicators that are closer to the listening observation: suppression, start loss, or strong reduction of speech content while the robot speaker is playing.
| Mic | N echo | Avg echo SNR dB | Avg echo active % | Avg echo envelope corr | Avg LUFS drop vs normal | Low loudness files | Low active files | Low speech-band files | Low SNR files | Low corr files | Long duration files | Risk files >=2 flags |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| HK 4-array v3.2 | 15 | 42 | 71.8 | 0.506 | 4 | 0 | 0 | 0 | 1 | 0 | 0 | 0 |
| CAE v1.2 | 21 | 31.5 | 65.6 | 0.421 | -4.4 | 7 | 4 | 1 | 3 | 6 | 1 | 8 |
9.1 Echo Files with High Risk Flags
| Mic | Distance | Trial | s | LUFS | LUFS drop | SNR | Active % | Speech % | Env corr | Leading ms | Dropout | Risk flags | Audio path |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| CAE v1.2 | 1m | 10 | 3 | -31.67 | -11.56 | 36.06 | 39.13 | 6.5 | 0.49 | 0 | 20.25 | 4 | micanhthao/micmoi/1m/10_echo.wav |
| CAE v1.2 | 1m | 11 | 32 | -24.11 | -4 | 35.25 | 38.04 | 92.24 | 0.16 | 90 | 14.94 | 4 | micanhthao/micmoi/1m/11_echo.wav |
| CAE v1.2 | 1m | 1 | 5 | -29.16 | -9.05 | 29.02 | 37.07 | 91.22 | 0.23 | 0 | 15.99 | 3 | micanhthao/micmoi/1m/1_echo.wav |
| CAE v1.2 | 2m | 4 | 5 | -26.53 | -6.08 | 27.79 | 80.96 | 91.91 | 0.34 | 0 | 3.09 | 2 | micanhthao/micmoi/2m/4_echo.wav |
| CAE v1.2 | 1m | 6 | 4 | -29.99 | -9.87 | 28.61 | 52.63 | 98.36 | 0.55 | 300 | 14.4 | 2 | micanhthao/micmoi/1m/6_echo.wav |
| CAE v1.2 | 2m | 1 | 5 | -28.55 | -8.1 | 29.13 | 58.52 | 86.47 | 0.28 | 0 | 7.58 | 2 | micanhthao/micmoi/2m/1_echo.wav |
| CAE v1.2 | 3m | 4 | 2 | -25.9 | -5.58 | 31.67 | 62.81 | 98.07 | 0.28 | 10 | 12.41 | 2 | micanhthao/micmoi/3m/4_echo.wav |
| CAE v1.2 | 1m | 3 | 4 | -23.87 | -3.76 | 20.91 | 99.25 | 90.31 | 0.23 | 0 | 0 | 2 | micanhthao/micmoi/1m/3_echo.wav |
| CAE v1.2 | 3m | 1 | 6 | -25.71 | -5.39 | 31.37 | 60.43 | 95.81 | 0.45 | 190 | 8.06 | 1 | micanhthao/micmoi/3m/1_echo.wav |
| CAE v1.2 | 1m | 8 | 3 | -20.92 | -0.8 | 37.91 | 36.45 | 96.39 | 0.86 | 70 | 18.85 | 1 | micanhthao/micmoi/1m/8_echo.wav |
| CAE v1.2 | 2m | 5 | 4 | -22.4 | -1.95 | 27.18 | 90.73 | 95.8 | 0.4 | 0 | 0.88 | 1 | micanhthao/micmoi/2m/5_echo.wav |
| HK mic 4 array linear v3.2 | 2m | 3 | 3 | -25.96 | 2.27 | 26.21 | 85.62 | 79.75 | 0.58 | 0 | 3.04 | 1 | michk_32/hk/2m/3_echo.wav |
| CAE v1.2 | 2m | 2 | 5 | -24.61 | -4.16 | 31.74 | 78.36 | 87.5 | 0.26 | 0 | 7.51 | 1 | micanhthao/micmoi/2m/2_echo.wav |
| HK mic 4 array linear v3.2 | 2m | 1 | 4 | -20.94 | 7.29 | 35.13 | 79.95 | 94.7 | 0.51 | 310 | 8.66 | 1 | michk_32/hk/2m/1_echo.wav |
10. Waveform and Energy Evidence for Echo Handling
The figures below plot waveform and RMS energy envelope for the real microphone and the simultaneous OBS/laptop recording in the same echo file. Lower real-microphone energy than OBS is not a true ERLE measurement by itself because gain and microphone placement also matter; however, it is useful visual evidence for suppression, missing segments, low active ratio, or speech preservation while the robot speaker is playing.
10.1 Representative Figures
HK 4-array v3.2 | 3m | trial 4 | path michk_32/hk/3m/4_echo.wav | risk flags 1
HK 4-array v3.2 | 2m | trial 2 | path michk_32/hk/2m/2_echo.wav | risk flags 1
CAE v1.2 | 1m | trial 11 | path micanhthao/micmoi/1m/11_echo.wav | risk flags 4
CAE v1.2 | 1m | trial 10 | path micanhthao/micmoi/1m/10_echo.wav | risk flags 4
CAE v1.2 | 1m | trial 1 | path micanhthao/micmoi/1m/1_echo.wav | risk flags 3
10.2 Full Waveform/Energy Figure Index for Echo Files
| Mic | Distance | Trial | Mic audio | OBS audio | LUFS | LUFS drop | SNR | Active % | Speech % | Risk flags | Wave/Energy figure |
|---|---|---|---|---|---|---|---|---|---|---|---|
| HK 4-array v3.2 | 1m | 1 | michk_32/hk/1m/1_echo.wav | michk_32/obs/1m/1_echo.wav | -23.68 | 3.27 | 36.37 | 74.25 | 87.66 | 0 | PNG |
| HK 4-array v3.2 | 1m | 2 | michk_32/hk/1m/2_echo.wav | michk_32/obs/1m/2_echo.wav | -20.72 | 6.24 | 59.9 | 64.88 | 86.28 | 0 | PNG |
| HK 4-array v3.2 | 1m | 3 | michk_32/hk/1m/3_echo.wav | michk_32/obs/1m/3_echo.wav | -21.15 | 5.8 | 37.66 | 72.34 | 79.71 | 0 | PNG |
| HK 4-array v3.2 | 1m | 4 | michk_32/hk/1m/4_echo.wav | michk_32/obs/1m/4_echo.wav | -20.86 | 6.1 | 31.65 | 85.21 | 87.49 | 0 | PNG |
| HK 4-array v3.2 | 1m | 5 | michk_32/hk/1m/5_echo.wav | michk_32/obs/1m/5_echo.wav | -21.28 | 5.68 | 39.71 | 68.92 | 81.22 | 0 | PNG |
| HK 4-array v3.2 | 2m | 1 | michk_32/hk/2m/1_echo.wav | michk_32/obs/2m/1_echo.wav | -20.94 | 7.29 | 35.13 | 79.95 | 94.7 | 1 | PNG |
| HK 4-array v3.2 | 2m | 2 | michk_32/hk/2m/2_echo.wav | michk_32/obs/2m/2_echo.wav | -25.89 | 2.33 | 57.35 | 62.66 | 94.13 | 1 | PNG |
| HK 4-array v3.2 | 2m | 3 | michk_32/hk/2m/3_echo.wav | michk_32/obs/2m/3_echo.wav | -25.96 | 2.27 | 26.21 | 85.62 | 79.75 | 1 | PNG |
| HK 4-array v3.2 | 2m | 4 | michk_32/hk/2m/4_echo.wav | michk_32/obs/2m/4_echo.wav | -23.36 | 4.87 | 29.85 | 82.94 | 87.22 | 0 | PNG |
| HK 4-array v3.2 | 2m | 5 | michk_32/hk/2m/5_echo.wav | michk_32/obs/2m/5_echo.wav | -25.67 | 2.56 | 39.85 | 69.28 | 85.45 | 0 | PNG |
| HK 4-array v3.2 | 3m | 1 | michk_32/hk/3m/1_echo.wav | michk_32/obs/3m/1_echo.wav | -25.75 | 1.85 | 36.97 | 74.69 | 93.97 | 0 | PNG |
| HK 4-array v3.2 | 3m | 2 | michk_32/hk/3m/2_echo.wav | michk_32/obs/3m/2_echo.wav | -23.16 | 4.44 | 36.26 | 71.43 | 93.63 | 0 | PNG |
| HK 4-array v3.2 | 3m | 3 | michk_32/hk/3m/3_echo.wav | michk_32/obs/3m/3_echo.wav | -25.62 | 1.98 | 53.78 | 62.91 | 90.92 | 1 | PNG |
| HK 4-array v3.2 | 3m | 4 | michk_32/hk/3m/4_echo.wav | michk_32/obs/3m/4_echo.wav | -25.26 | 2.33 | 57.57 | 58.92 | 91.08 | 1 | PNG |
| HK 4-array v3.2 | 3m | 5 | michk_32/hk/3m/5_echo.wav | michk_32/obs/3m/5_echo.wav | -24.69 | 2.9 | 51.94 | 63.53 | 86.27 | 0 | PNG |
| CAE v1.2 | 1m | 1 | micanhthao/micmoi/1m/1_echo.wav | micanhthao/obs/1m/1_echo.wav | -29.16 | -9.05 | 29.02 | 37.07 | 91.22 | 3 | PNG |
| CAE v1.2 | 1m | 2 | micanhthao/micmoi/1m/2_echo.wav | micanhthao/obs/1m/2_echo.wav | -22.76 | -2.65 | 34.35 | 55.09 | 93.44 | 0 | PNG |
| CAE v1.2 | 1m | 3 | micanhthao/micmoi/1m/3_echo.wav | micanhthao/obs/1m/3_echo.wav | -23.87 | -3.76 | 20.91 | 99.25 | 90.31 | 2 | PNG |
| CAE v1.2 | 1m | 4 | micanhthao/micmoi/1m/4_echo.wav | micanhthao/obs/1m/4_echo.wav | -21.73 | -1.61 | 36.07 | 72.29 | 93.95 | 0 | PNG |
| CAE v1.2 | 1m | 5 | micanhthao/micmoi/1m/5_echo.wav | micanhthao/obs/1m/5_echo.wav | -24.89 | -4.77 | 32.85 | 63.16 | 96.52 | 0 | PNG |
| CAE v1.2 | 1m | 6 | micanhthao/micmoi/1m/6_echo.wav | micanhthao/obs/1m/6_echo.wav | -29.99 | -9.87 | 28.61 | 52.63 | 98.36 | 2 | PNG |
| CAE v1.2 | 1m | 7 | micanhthao/micmoi/1m/7_echo.wav | micanhthao/obs/1m/7_echo.wav | -22.96 | -2.85 | 28.61 | 86.72 | 90.73 | 0 | PNG |
| CAE v1.2 | 1m | 8 | micanhthao/micmoi/1m/8_echo.wav | micanhthao/obs/1m/8_echo.wav | -20.92 | -0.8 | 37.91 | 36.45 | 96.39 | 1 | PNG |
| CAE v1.2 | 1m | 9 | micanhthao/micmoi/1m/9_echo.wav | micanhthao/obs/1m/9_echo.wav | -22.51 | -2.4 | 31.47 | 81.61 | 91.36 | 0 | PNG |
| CAE v1.2 | 1m | 10 | micanhthao/micmoi/1m/10_echo.wav | micanhthao/obs/1m/10_echo.wav | -31.67 | -11.56 | 36.06 | 39.13 | 6.5 | 4 | PNG |
| CAE v1.2 | 1m | 11 | micanhthao/micmoi/1m/11_echo.wav | micanhthao/obs/1m/11_echo.wav | -24.11 | -4 | 35.25 | 38.04 | 92.24 | 4 | PNG |
| CAE v1.2 | 2m | 1 | micanhthao/micmoi/2m/1_echo.wav | micanhthao/obs/2m/1_echo.wav | -28.55 | -8.1 | 29.13 | 58.52 | 86.47 | 2 | PNG |
| CAE v1.2 | 2m | 2 | micanhthao/micmoi/2m/2_echo.wav | micanhthao/obs/2m/2_echo.wav | -24.61 | -4.16 | 31.74 | 78.36 | 87.5 | 1 | PNG |
| CAE v1.2 | 2m | 3 | micanhthao/micmoi/2m/3_echo.wav | micanhthao/obs/2m/3_echo.wav | -22.52 | -2.07 | 30.48 | 81.45 | 87.52 | 0 | PNG |
| CAE v1.2 | 2m | 4 | micanhthao/micmoi/2m/4_echo.wav | micanhthao/obs/2m/4_echo.wav | -26.53 | -6.08 | 27.79 | 80.96 | 91.91 | 2 | PNG |
| CAE v1.2 | 2m | 5 | micanhthao/micmoi/2m/5_echo.wav | micanhthao/obs/2m/5_echo.wav | -22.4 | -1.95 | 27.18 | 90.73 | 95.8 | 1 | PNG |
| CAE v1.2 | 3m | 1 | micanhthao/micmoi/3m/1_echo.wav | micanhthao/obs/3m/1_echo.wav | -25.71 | -5.39 | 31.37 | 60.43 | 95.81 | 1 | PNG |
| CAE v1.2 | 3m | 2 | micanhthao/micmoi/3m/2_echo.wav | micanhthao/obs/3m/2_echo.wav | -21.84 | -1.52 | 33.93 | 73.93 | 96.39 | 0 | PNG |
| CAE v1.2 | 3m | 3 | micanhthao/micmoi/3m/3_echo.wav | micanhthao/obs/3m/3_echo.wav | -23.22 | -2.9 | 33.99 | 58.86 | 97.6 | 0 | PNG |
| CAE v1.2 | 3m | 4 | micanhthao/micmoi/3m/4_echo.wav | micanhthao/obs/3m/4_echo.wav | -25.9 | -5.58 | 31.67 | 62.81 | 98.07 | 2 | PNG |
| CAE v1.2 | 3m | 5 | micanhthao/micmoi/3m/5_echo.wav | micanhthao/obs/3m/5_echo.wav | -22.51 | -2.19 | 32.35 | 70.68 | 97.87 | 0 | PNG |
11. Files with the Strongest Energy Dropout / Silence Evidence
| Mic | Distance | Mode | Trial | s | LUFS | SNR | Active % | Leading ms | Longest silence ms | Dropout | Audio path |
|---|---|---|---|---|---|---|---|---|---|---|---|
| HK mic 4 array linear v3.2 | 1m | Normal | 5 | 4 | -27.77 | 59.58 | 44.61 | 1470 | 1470 | 38.51 | michk_32/hk/1m/5.wav |
| HK mic 4 array linear v3.2 | 3m | Echo | 4 | 5 | -25.26 | 57.57 | 58.92 | 1090 | 1090 | 24.15 | michk_32/hk/3m/4_echo.wav |
| CAE v1.2 | 1m | Normal | 1 | 5 | -19.24 | 41.01 | 49.1 | 750 | 950 | 26.36 | micanhthao/micmoi/1m/1.wav |
| CAE v1.2 | 3m | Normal | 8 | 6 | -20.46 | 39.2 | 62.44 | 680 | 920 | 19.51 | micanhthao/micmoi/3m/8.wav |
| CAE v1.2 | 1m | Normal | 4 | 3 | -19.94 | 39.94 | 49.5 | 610 | 710 | 28.31 | micanhthao/micmoi/1m/4.wav |
| CAE v1.2 | 1m | Normal | 2 | 4 | -20.59 | 39.99 | 47.12 | 600 | 620 | 24.76 | micanhthao/micmoi/1m/2.wav |
| CAE v1.2 | 2m | Normal | 2 | 4 | -19.65 | 40.31 | 59.4 | 600 | 690 | 21.25 | micanhthao/micmoi/2m/2.wav |
| HK mic 4 array linear v3.2 | 1m | Normal | 1 | 6 | -26.17 | 60.41 | 68.78 | 540 | 1210 | 19.33 | michk_32/hk/1m/1.wav |
| HK mic 4 array linear v3.2 | 3m | Normal | 4 | 5 | -27.2 | 59.51 | 63.93 | 540 | 920 | 19.49 | michk_32/hk/3m/4.wav |
| HK mic 4 array linear v3.2 | 1m | Normal | 3 | 5 | -27.19 | 59.37 | 64.13 | 500 | 1040 | 20.94 | michk_32/hk/1m/3.wav |
| CAE v1.2 | 3m | Normal | 7 | 6 | -20.54 | 38.94 | 58.76 | 470 | 1030 | 20.56 | micanhthao/micmoi/3m/7.wav |
| HK mic 4 array linear v3.2 | 3m | Normal | 3 | 4 | -26.86 | 54.53 | 72.68 | 520 | 520 | 15.41 | michk_32/hk/3m/3.wav |
12. Technical and Operational Assessment
HK mic 4 array linear v3.2
- No echo: usable capture quality at 1 m, 2 m, and 3 m.
- Echo: less information loss than CAE, although mild distortion is still expected when AEC / echo suppression is active.
- Deployment: compact, fewer external accessories, and built-in tuning software, making it more suitable for repeatable robot integration.
CAE v1.2
- No echo: usable capture quality, matching the subjective observation that both microphones are acceptable without robot-speaker playback.
- Echo: higher risk of speech suppression / information loss in some files. In this dataset, no file is fully muted by the absolute-energy threshold, but multiple CAE echo files show loudness drops, low active ratio, abnormal speech-band energy, or low correlation with the OBS reference; these are listed in the echo risk table.
- Deployment: requires an external sound card and additional wiring/hardware, increasing integration complexity.
13. Generated Figures
figures/snr_hk_vs_cae.png: SNR by distance and mode.figures/lufs_hk_vs_cae.png: LUFS by distance and mode.figures/dropout_hk_vs_cae.png: dropout score by distance and mode.figures/leading_silence_hk_vs_cae.png: leading audio loss by distance and mode.figures/pair_rms_delta_mic_obs.png: real microphone RMS delta against laptop reference.figures/pair_active_delta_mic_obs.png: real microphone active-ratio delta against laptop reference.figures/pair_envelope_corr_mic_obs.png: real microphone/laptop envelope correlation.figures/echo_wave_energy/*.png: waveform and RMS energy envelope for each echo pair.
14. Per-File Real-Microphone Details
| Mic | Distance | Mode | Trial | s | LUFS | RMS | Peak | SNR | Active % | Leading ms | Longest silence ms | Dropout | Speech % | Echo idx | Clip % | Audio path |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| HK mic 4 array linear v3.2 | 1m | Normal | 1 | 6 | -26.17 | -26.48 | -9.29 | 60.41 | 68.78 | 540 | 1210 | 19.33 | 58.68 | 0.9 | 0 | michk_32/hk/1m/1.wav |
| HK mic 4 array linear v3.2 | 1m | Normal | 2 | 4 | -26.64 | -26.8 | -8.45 | 41.35 | 77.19 | 0 | 780 | 13.68 | 62.38 | 0.9 | 0 | michk_32/hk/1m/2.wav |
| HK mic 4 array linear v3.2 | 1m | Normal | 3 | 5 | -27.19 | -28.01 | -10.04 | 59.37 | 64.13 | 500 | 1040 | 20.94 | 59.85 | 0.89 | 0 | michk_32/hk/1m/3.wav |
| HK mic 4 array linear v3.2 | 1m | Normal | 4 | 3 | -27.01 | -27.55 | -10.03 | 39.91 | 74.58 | 50 | 530 | 13.68 | 75.2 | 0.77 | 0 | michk_32/hk/1m/4.wav |
| HK mic 4 array linear v3.2 | 1m | Normal | 5 | 4 | -27.77 | -30.13 | -7.82 | 59.58 | 44.61 | 1470 | 1470 | 38.51 | 83.53 | 0.88 | 0 | michk_32/hk/1m/5.wav |
| CAE v1.2 | 1m | Normal | 1 | 5 | -19.24 | -21.18 | -1.81 | 41.01 | 49.1 | 750 | 950 | 26.36 | 45.97 | 0.92 | 0 | micanhthao/micmoi/1m/1.wav |
| CAE v1.2 | 1m | Normal | 2 | 4 | -20.59 | -22.45 | -1.83 | 39.99 | 47.12 | 600 | 620 | 24.76 | 56.36 | 0.92 | 0 | micanhthao/micmoi/1m/2.wav |
| CAE v1.2 | 1m | Normal | 3 | 4 | -19.93 | -21.07 | -5.34 | 39.76 | 61.15 | 180 | 780 | 19.32 | 64.13 | 0.91 | 0 | micanhthao/micmoi/1m/3.wav |
| CAE v1.2 | 1m | Normal | 4 | 3 | -19.94 | -21.86 | -4.55 | 39.94 | 49.5 | 610 | 710 | 28.31 | 56.2 | 0.91 | 0 | micanhthao/micmoi/1m/4.wav |
| CAE v1.2 | 1m | Normal | 5 | 4 | -20.11 | -20.62 | -3.9 | 39.61 | 69.92 | 0 | 820 | 16.49 | 50.83 | 0.91 | 0 | micanhthao/micmoi/1m/5.wav |
| CAE v1.2 | 1m | Normal | 6 | 4 | -20.54 | -22.07 | -4.55 | 39.72 | 55.14 | 340 | 820 | 22.84 | 57.95 | 0.91 | 0 | micanhthao/micmoi/1m/6.wav |
| CAE v1.2 | 1m | Normal | 7 | 3 | -20.16 | -20.3 | -5.34 | 38.19 | 73.91 | 380 | 380 | 13.19 | 38.66 | 0.84 | 0 | micanhthao/micmoi/1m/7.wav |
| CAE v1.2 | 1m | Normal | 8 | 3 | -20.21 | -20.1 | -4.65 | 35.98 | 80.6 | 0 | 160 | 6.67 | 52.75 | 0.86 | 0 | micanhthao/micmoi/1m/8.wav |
| CAE v1.2 | 1m | Normal | 9 | 5 | -20.18 | -21.74 | -4.34 | 39.8 | 57.11 | 360 | 690 | 19.61 | 60.04 | 0.9 | 0 | micanhthao/micmoi/1m/9.wav |
| CAE v1.2 | 1m | Normal | 10 | 4 | -20.23 | -21.69 | -4.52 | 39.35 | 60.15 | 40 | 1230 | 23.27 | 41.98 | 0.89 | 0 | micanhthao/micmoi/1m/10.wav |
| HK mic 4 array linear v3.2 | 2m | Normal | 1 | 4 | -27.87 | -28.61 | -12.77 | 57.69 | 67.92 | 200 | 950 | 19.69 | 73.42 | 0.87 | 0 | michk_32/hk/2m/1.wav |
| HK mic 4 array linear v3.2 | 2m | Normal | 2 | 4 | -28.1 | -28.66 | -11.23 | 58.14 | 68.67 | 40 | 1010 | 18.97 | 80.25 | 0.85 | 0 | michk_32/hk/2m/2.wav |
| HK mic 4 array linear v3.2 | 2m | Normal | 3 | 4 | -27.41 | -28.36 | -8.35 | 59.13 | 57.89 | 240 | 1050 | 22.69 | 79.79 | 0.85 | 0 | michk_32/hk/2m/3.wav |
| HK mic 4 array linear v3.2 | 2m | Normal | 4 | 5 | -26.55 | -26.66 | -9.84 | 43.33 | 75.55 | 40 | 960 | 14.19 | 90.43 | 0.81 | 0 | michk_32/hk/2m/4.wav |
| HK mic 4 array linear v3.2 | 2m | Normal | 5 | 3 | -31.23 | -31.94 | -10.41 | 55.15 | 54.85 | 0 | 1150 | 27.63 | 78.8 | 0.89 | 0 | michk_32/hk/2m/5.wav |
| CAE v1.2 | 2m | Normal | 1 | 4 | -20.65 | -22.49 | -2.22 | 39.23 | 53.88 | 280 | 610 | 20.79 | 57.14 | 0.9 | 0 | micanhthao/micmoi/2m/1.wav |
| CAE v1.2 | 2m | Normal | 2 | 4 | -19.65 | -20.92 | -4.71 | 40.31 | 59.4 | 600 | 690 | 21.25 | 71.3 | 0.89 | 0 | micanhthao/micmoi/2m/2.wav |
| CAE v1.2 | 2m | Normal | 3 | 4 | -18.91 | -20.54 | -2.78 | 41.12 | 52.63 | 190 | 1310 | 27.36 | 58 | 0.9 | 0 | micanhthao/micmoi/2m/3.wav |
| CAE v1.2 | 2m | Normal | 4 | 4.5 | -21.34 | -22.99 | -5.06 | 38.26 | 56.79 | 0 | 1300 | 24.01 | 47.78 | 0.9 | 0 | micanhthao/micmoi/2m/4.wav |
| CAE v1.2 | 2m | Normal | 5 | 4 | -19.7 | -20.12 | -4.05 | 39.14 | 75.69 | 340 | 410 | 12.93 | 52.63 | 0.77 | 0 | micanhthao/micmoi/2m/5.wav |
| CAE v1.2 | 2m | Normal | 6 | 4.5 | -20.54 | -20.73 | -3.07 | 36.75 | 78.84 | 90 | 490 | 9.68 | 55.19 | 0.76 | 0 | micanhthao/micmoi/2m/6.wav |
| CAE v1.2 | 2m | Normal | 7 | 4 | -19.77 | -20.68 | -2.67 | 39.88 | 64.41 | 340 | 730 | 18.37 | 53.24 | 0.84 | 0 | micanhthao/micmoi/2m/7.wav |
| CAE v1.2 | 2m | Normal | 8 | 4 | -20.52 | -22.21 | -3.3 | 39.56 | 53.88 | 150 | 540 | 18.38 | 60.62 | 0.88 | 0 | micanhthao/micmoi/2m/8.wav |
| CAE v1.2 | 2m | Normal | 9 | 4 | -20.59 | -21.44 | -4.68 | 38.47 | 67.67 | 310 | 620 | 16.47 | 77.2 | 0.83 | 0 | micanhthao/micmoi/2m/9.wav |
| CAE v1.2 | 2m | Normal | 10 | 5 | -22.81 | -23.68 | -7.06 | 35.51 | 71.34 | 0 | 280 | 9.26 | 66.53 | 0.78 | 0 | micanhthao/micmoi/2m/10.wav |
| HK mic 4 array linear v3.2 | 3m | Normal | 1 | 4 | -27.09 | -27.71 | -10.64 | 57.8 | 73.43 | 0 | 930 | 16.4 | 86.81 | 0.91 | 0 | michk_32/hk/3m/1.wav |
| HK mic 4 array linear v3.2 | 3m | Normal | 2 | 4 | -28.17 | -28.73 | -12.17 | 58.66 | 68.92 | 220 | 900 | 18.82 | 89.04 | 0.84 | 0 | michk_32/hk/3m/2.wav |
| HK mic 4 array linear v3.2 | 3m | Normal | 3 | 4 | -26.86 | -27.27 | -9.36 | 54.53 | 72.68 | 520 | 520 | 15.41 | 93.71 | 0.81 | 0 | michk_32/hk/3m/3.wav |
| HK mic 4 array linear v3.2 | 3m | Normal | 4 | 5 | -27.2 | -27.85 | -10.54 | 59.51 | 63.93 | 540 | 920 | 19.49 | 93.4 | 0.86 | 0 | michk_32/hk/3m/4.wav |
| HK mic 4 array linear v3.2 | 3m | Normal | 5 | 4 | -28.67 | -29.55 | -13 | 57.02 | 67.92 | 0 | 1160 | 20.44 | 92.9 | 0.9 | 0 | michk_32/hk/3m/5.wav |
| CAE v1.2 | 3m | Normal | 1 | 3.5 | -20.33 | -21.6 | -4.02 | 38.76 | 63.04 | 310 | 590 | 17.42 | 43.15 | 0.87 | 0 | micanhthao/micmoi/3m/1.wav |
| CAE v1.2 | 3m | Normal | 2 | 3.5 | -19.34 | -21.1 | -4.64 | 39.86 | 58.45 | 0 | 1210 | 24.77 | 59.88 | 0.91 | 0 | micanhthao/micmoi/3m/2.wav |
| CAE v1.2 | 3m | Normal | 3 | 5 | -19.94 | -21.85 | -4.62 | 39.66 | 52.71 | 380 | 510 | 19.34 | 41.81 | 0.92 | 0 | micanhthao/micmoi/3m/3.wav |
| CAE v1.2 | 3m | Normal | 4 | 5 | -20.88 | -22.33 | -2.83 | 38.61 | 61.72 | 120 | 750 | 17.03 | 49.35 | 0.88 | 0 | micanhthao/micmoi/3m/4.wav |
| CAE v1.2 | 3m | Normal | 5 | 3 | -18.49 | -19.43 | -4.8 | 40.82 | 66.89 | 0 | 340 | 12.41 | 52.22 | 0.87 | 0 | micanhthao/micmoi/3m/5.wav |
| CAE v1.2 | 3m | Normal | 6 | 4 | -20.45 | -22.72 | -5.59 | 39.66 | 48.37 | 80 | 1710 | 32.42 | 50.15 | 0.9 | 0 | micanhthao/micmoi/3m/6.wav |
| CAE v1.2 | 3m | Normal | 7 | 6 | -20.54 | -22.12 | -2.44 | 38.94 | 58.76 | 470 | 1030 | 20.56 | 53.76 | 0.91 | 0 | micanhthao/micmoi/3m/7.wav |
| CAE v1.2 | 3m | Normal | 8 | 6 | -20.46 | -21.71 | -4.44 | 39.2 | 62.44 | 680 | 920 | 19.51 | 53.11 | 0.9 | 0 | micanhthao/micmoi/3m/8.wav |
| CAE v1.2 | 3m | Normal | 9 | 6 | -22.2 | -24.36 | -2.32 | 37.53 | 52.09 | 330 | 1130 | 22.84 | 52.8 | 0.88 | 0 | micanhthao/micmoi/3m/9.wav |
| CAE v1.2 | 3m | Normal | 10 | 7 | -20.54 | -21.86 | -4.05 | 39.06 | 59.37 | 0 | 1430 | 18.73 | 53.73 | 0.93 | 0 | micanhthao/micmoi/3m/10.wav |
| HK mic 4 array linear v3.2 | 1m | Echo | 1 | 3 | -23.68 | -24.89 | -2.43 | 36.37 | 74.25 | 0 | 140 | 7.84 | 87.66 | 0.59 | 0 | michk_32/hk/1m/1_echo.wav |
| HK mic 4 array linear v3.2 | 1m | Echo | 2 | 3 | -20.72 | -22.56 | -2.35 | 59.9 | 64.88 | 0 | 590 | 17.56 | 86.28 | 0.79 | 0 | michk_32/hk/1m/2_echo.wav |
| HK mic 4 array linear v3.2 | 1m | Echo | 3 | 5 | -21.15 | -23.41 | -2.17 | 37.66 | 72.34 | 0 | 130 | 7.43 | 79.71 | 0.59 | 0 | michk_32/hk/1m/3_echo.wav |
| HK mic 4 array linear v3.2 | 1m | Echo | 4 | 4 | -20.86 | -21.7 | -2.86 | 31.65 | 85.21 | 160 | 160 | 5.28 | 87.49 | 0.53 | 0 | michk_32/hk/1m/4_echo.wav |
| HK mic 4 array linear v3.2 | 1m | Echo | 5 | 4 | -21.28 | -22.41 | -3.1 | 39.71 | 68.92 | 0 | 460 | 11.4 | 81.22 | 0.67 | 0 | michk_32/hk/1m/5_echo.wav |
| CAE v1.2 | 1m | Echo | 1 | 5 | -29.16 | -34.82 | -9.5 | 29.02 | 37.07 | 0 | 530 | 15.99 | 91.22 | 0.72 | 0 | micanhthao/micmoi/1m/1_echo.wav |
| CAE v1.2 | 1m | Echo | 2 | 6 | -22.76 | -25.77 | -6.29 | 34.35 | 55.09 | 0 | 180 | 7.89 | 93.44 | 0.79 | 0 | micanhthao/micmoi/1m/2_echo.wav |
| CAE v1.2 | 1m | Echo | 3 | 4 | -23.87 | -23.74 | -7 | 20.91 | 99.25 | 0 | 0 | 0 | 90.31 | 0.37 | 0 | micanhthao/micmoi/1m/3_echo.wav |
| CAE v1.2 | 1m | Echo | 4 | 6 | -21.73 | -23.2 | -4.88 | 36.07 | 72.29 | 0 | 630 | 10.34 | 93.95 | 0.84 | 0 | micanhthao/micmoi/1m/4_echo.wav |
| CAE v1.2 | 1m | Echo | 5 | 4 | -24.89 | -27.25 | -5.22 | 32.85 | 63.16 | 0 | 330 | 10.18 | 96.52 | 0.83 | 0 | micanhthao/micmoi/1m/5_echo.wav |
| CAE v1.2 | 1m | Echo | 6 | 4 | -29.99 | -32.27 | -9.79 | 28.61 | 52.63 | 300 | 300 | 14.4 | 98.36 | 0.79 | 0 | micanhthao/micmoi/1m/6_echo.wav |
| CAE v1.2 | 1m | Echo | 7 | 4 | -22.96 | -24.05 | -6.85 | 28.61 | 86.72 | 0 | 100 | 2.72 | 90.73 | 0.77 | 0 | micanhthao/micmoi/1m/7_echo.wav |
| CAE v1.2 | 1m | Echo | 8 | 3 | -20.92 | -25.21 | -4.47 | 37.91 | 36.45 | 70 | 390 | 18.85 | 96.39 | 0.87 | 0 | micanhthao/micmoi/1m/8_echo.wav |
| CAE v1.2 | 1m | Echo | 9 | 3 | -22.51 | -24.1 | -2.79 | 31.47 | 81.61 | 130 | 130 | 5.43 | 91.36 | 0.77 | 0 | micanhthao/micmoi/1m/9_echo.wav |
| CAE v1.2 | 1m | Echo | 10 | 3 | -31.67 | -27.81 | -4.62 | 36.06 | 39.13 | 0 | 520 | 20.25 | 6.5 | 0.87 | 0 | micanhthao/micmoi/1m/10_echo.wav |
| CAE v1.2 | 1m | Echo | 11 | 32 | -24.11 | -28.98 | -3.44 | 35.25 | 38.04 | 90 | 1280 | 14.94 | 92.24 | 0.81 | 0 | micanhthao/micmoi/1m/11_echo.wav |
| HK mic 4 array linear v3.2 | 2m | Echo | 1 | 4 | -20.94 | -22.29 | -4.07 | 35.13 | 79.95 | 310 | 310 | 8.66 | 94.7 | 0.68 | 0 | michk_32/hk/2m/1_echo.wav |
| HK mic 4 array linear v3.2 | 2m | Echo | 2 | 4 | -25.89 | -26.26 | -6.76 | 57.35 | 62.66 | 0 | 1200 | 22.11 | 94.13 | 0.83 | 0 | michk_32/hk/2m/2_echo.wav |
| HK mic 4 array linear v3.2 | 2m | Echo | 3 | 3 | -25.96 | -26.14 | -5.96 | 26.21 | 85.62 | 0 | 50 | 3.04 | 79.75 | 0.41 | 0 | michk_32/hk/2m/3_echo.wav |
| HK mic 4 array linear v3.2 | 2m | Echo | 4 | 3 | -23.36 | -24.33 | -4.91 | 29.85 | 82.94 | 0 | 70 | 3.63 | 87.22 | 0.52 | 0 | michk_32/hk/2m/4_echo.wav |
| HK mic 4 array linear v3.2 | 2m | Echo | 5 | 6 | -25.67 | -26.61 | -3.58 | 39.85 | 69.28 | 250 | 350 | 11.59 | 85.45 | 0.64 | 0 | michk_32/hk/2m/5_echo.wav |
| CAE v1.2 | 2m | Echo | 1 | 5 | -28.55 | -31.05 | -11.53 | 29.13 | 58.52 | 0 | 130 | 7.58 | 86.47 | 0.82 | 0 | micanhthao/micmoi/2m/1_echo.wav |
| CAE v1.2 | 2m | Echo | 2 | 5 | -24.61 | -24.96 | -4.76 | 31.74 | 78.36 | 0 | 420 | 7.51 | 87.5 | 0.81 | 0 | micanhthao/micmoi/2m/2_echo.wav |
| CAE v1.2 | 2m | Echo | 3 | 4 | -22.52 | -24.11 | -5.06 | 30.48 | 81.45 | 0 | 80 | 3.07 | 87.52 | 0.81 | 0 | micanhthao/micmoi/2m/3_echo.wav |
| CAE v1.2 | 2m | Echo | 4 | 5 | -26.53 | -27.5 | -7.59 | 27.79 | 80.96 | 0 | 90 | 3.09 | 91.91 | 0.74 | 0 | micanhthao/micmoi/2m/4_echo.wav |
| CAE v1.2 | 2m | Echo | 5 | 4 | -22.4 | -23.94 | -5.68 | 27.18 | 90.73 | 0 | 40 | 0.88 | 95.8 | 0.67 | 0 | micanhthao/micmoi/2m/5_echo.wav |
| HK mic 4 array linear v3.2 | 3m | Echo | 1 | 4 | -25.75 | -26 | -3.87 | 36.97 | 74.69 | 0 | 320 | 9.56 | 93.97 | 0.59 | 0 | michk_32/hk/3m/1_echo.wav |
| HK mic 4 array linear v3.2 | 3m | Echo | 2 | 4 | -23.16 | -24.82 | -4.08 | 36.26 | 71.43 | 0 | 160 | 7.98 | 93.63 | 0.61 | 0 | michk_32/hk/3m/2_echo.wav |
| HK mic 4 array linear v3.2 | 3m | Echo | 3 | 4 | -25.62 | -28.15 | -5.35 | 53.78 | 62.91 | 0 | 960 | 20 | 90.92 | 0.86 | 0 | michk_32/hk/3m/3_echo.wav |
| HK mic 4 array linear v3.2 | 3m | Echo | 4 | 5 | -25.26 | -26.91 | -4.16 | 57.57 | 58.92 | 1090 | 1090 | 24.15 | 91.08 | 0.85 | 0 | michk_32/hk/3m/4_echo.wav |
| HK mic 4 array linear v3.2 | 3m | Echo | 5 | 5 | -24.69 | -26.7 | -8.09 | 51.94 | 63.53 | 50 | 350 | 14.16 | 86.27 | 0.7 | 0 | michk_32/hk/3m/5_echo.wav |
| CAE v1.2 | 3m | Echo | 1 | 6 | -25.71 | -28.15 | -6.22 | 31.37 | 60.43 | 190 | 250 | 8.06 | 95.81 | 0.75 | 0 | micanhthao/micmoi/3m/1_echo.wav |
| CAE v1.2 | 3m | Echo | 2 | 4 | -21.84 | -23.72 | -4.81 | 33.93 | 73.93 | 120 | 290 | 7.97 | 96.39 | 0.8 | 0 | micanhthao/micmoi/3m/2_echo.wav |
| CAE v1.2 | 3m | Echo | 3 | 3 | -23.22 | -26.19 | -7 | 33.99 | 58.86 | 0 | 290 | 10.65 | 97.6 | 0.82 | 0 | micanhthao/micmoi/3m/3_echo.wav |
| CAE v1.2 | 3m | Echo | 4 | 2 | -25.9 | -28.71 | -11.46 | 31.67 | 62.81 | 10 | 260 | 12.41 | 98.07 | 0.72 | 0 | micanhthao/micmoi/3m/4_echo.wav |
| CAE v1.2 | 3m | Echo | 5 | 4 | -22.51 | -26.3 | -6.35 | 32.35 | 70.68 | 0 | 330 | 8.16 | 97.87 | 0.77 | 0 | micanhthao/micmoi/3m/5_echo.wav |
15. Limitations
This report measures the available recordings only. Because there is no separated far-end / speaker reference signal, true ERLE is not computed; echo and dropout metrics are heuristics for relative comparison under this test set.