From db8bc2c71d76ddbe475eb291b8bcaab09ca20d02 Mon Sep 17 00:00:00 2001 From: Alicia <115451386+congw729@users.noreply.github.com> Date: Thu, 18 Jun 2026 16:56:13 +0800 Subject: [PATCH] Update VoxCPM and Qwen3TTS perf baseline. (#4521) Signed-off-by: Alicia <115451386+congw729@users.noreply.github.com> --- tests/dfx/perf/tests/test_tts.json | 42 +++++++++++++------------- tests/dfx/perf/tests/test_voxcpm2.json | 26 ++++++++-------- 2 files changed, 34 insertions(+), 34 deletions(-) diff --git a/tests/dfx/perf/tests/test_tts.json b/tests/dfx/perf/tests/test_tts.json index ba1f95261..1ba273a2c 100644 --- a/tests/dfx/perf/tests/test_tts.json +++ b/tests/dfx/perf/tests/test_tts.json @@ -27,10 +27,10 @@ "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration", "baseline": { "median_audio_ttfp_ms": [ - 122.4504 + 139.9434 ], "median_audio_rtf": [ - 0.1472 + 0.1683 ] } }, @@ -55,17 +55,17 @@ "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration", "baseline": { "median_audio_ttfp_ms": [ - 199.1197, + 227.5653, 647.1272, 4777.2714 ], "median_audio_rtf": [ - 0.2899, + 0.3313, 0.3642, 1.2282 ], "audio_throughput": [ - 20.5634, + 23.501, 40.634, 43.0758 ] @@ -129,10 +129,10 @@ "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration", "baseline": { "median_audio_ttfp_ms": [ - 43.5669 + 49.7908 ], "median_audio_rtf": [ - 0.1247 + 0.1425 ] } }, @@ -162,17 +162,17 @@ "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration", "baseline": { "median_audio_ttfp_ms": [ - 65.9174, + 75.3341, 713.7727, 5941.7179 ], "median_audio_rtf": [ - 0.1626, + 0.1859, 0.3148, 1.1662 ], "audio_throughput": [ - 31.9068, + 36.465, 49.6268, 48.3091 ] @@ -228,13 +228,13 @@ "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration", "baseline": { "median_audio_ttfp_ms": [ - 49.6105 + 56.6977 ], "median_audio_rtf": [ - 0.1361 + 0.1555 ], "audio_throughput": [ - 10.0881 + 11.5292 ] } }, @@ -259,10 +259,10 @@ "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration", "baseline": { "median_audio_ttfp_ms": [ - 45.5391 + 52.0447 ], "median_audio_rtf": [ - 0.1271 + 0.1452 ] } }, @@ -291,17 +291,17 @@ "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration", "baseline": { "median_audio_ttfp_ms": [ - 66.1471, + 75.5967, 587.5877, 4960.3726 ], "median_audio_rtf": [ - 0.1637, + 0.1871, 0.3143, 1.1703 ], "audio_throughput": [ - 35.1923, + 40.2198, 48.9046, 48.0505 ] @@ -328,13 +328,13 @@ "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration", "baseline": { "median_audio_ttfp_ms": [ - 49.3466 + 56.3961 ], "median_audio_rtf": [ - 0.1349 + 0.1542 ], "audio_throughput": [ - 9 + 9.6672 ] } } diff --git a/tests/dfx/perf/tests/test_voxcpm2.json b/tests/dfx/perf/tests/test_voxcpm2.json index 0aaca446f..beb40d5d3 100644 --- a/tests/dfx/perf/tests/test_voxcpm2.json +++ b/tests/dfx/perf/tests/test_voxcpm2.json @@ -23,10 +23,10 @@ "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration", "baseline": { "median_audio_ttfp_ms": [ - 113.9351 + 130.2115 ], "median_audio_rtf": [ - 0.0427 + 0.0488 ] } }, @@ -47,13 +47,13 @@ "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration", "baseline": { "median_audio_ttfp_ms": [ - 415.9187 + 475.3357 ], "median_audio_rtf": [ - 0.1462 + 0.1671 ], "audio_throughput": [ - 20.0128 + 22.8717 ] } }, @@ -100,10 +100,10 @@ "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration", "baseline": { "median_audio_ttfp_ms": [ - 66.2748 + 75.7427 ], "median_audio_rtf": [ - 0.0366 + 0.0419 ] } }, @@ -127,13 +127,13 @@ "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration", "baseline": { "median_audio_ttfp_ms": [ - 307.6373 + 351.5855 ], "median_audio_rtf": [ - 0.1056 + 0.1207 ], "audio_throughput": [ - 64.5428 + 73.7632 ] } }, @@ -157,13 +157,13 @@ "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration", "baseline": { "median_audio_ttfp_ms": [ - 70.0958 + 80.1095 ], "median_audio_rtf": [ - 0.0424 + 0.0485 ], "audio_throughput": [ - 16.075 + 18.3715 ] } },