From 565f7475c7c9a6768158ec6b08b42698c0a88639 Mon Sep 17 00:00:00 2001
From: tzhouam <tzhouam@connect.ust.hk>
Date: Wed, 9 Sep 2026 16:45:25 +0800
Subject: [PATCH] edge results: codec stream cadence check 2 (chain6, server
 per-step outputs confirmed)

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
---
 .../20260909-163326/client_0.json             | 60 ++++++++++++++++
 .../20260909-163326/client_1.json             | 60 ++++++++++++++++
 .../20260909-163326/client_2.json             | 60 ++++++++++++++++
 .../20260909-163326/client_3.json             | 60 ++++++++++++++++
 .../codec_stream/20260909-163326/summary.json | 71 +++++++++++++++++++
 5 files changed, 311 insertions(+)
 create mode 100644 benchmarks/tts/edge_results/codec_stream/20260909-163326/client_0.json
 create mode 100644 benchmarks/tts/edge_results/codec_stream/20260909-163326/client_1.json
 create mode 100644 benchmarks/tts/edge_results/codec_stream/20260909-163326/client_2.json
 create mode 100644 benchmarks/tts/edge_results/codec_stream/20260909-163326/client_3.json
 create mode 100644 benchmarks/tts/edge_results/codec_stream/20260909-163326/summary.json

diff --git a/benchmarks/tts/edge_results/codec_stream/20260909-163326/client_0.json b/benchmarks/tts/edge_results/codec_stream/20260909-163326/client_0.json
new file mode 100644
index 000000000..a22f3097e
--- /dev/null
+++ b/benchmarks/tts/edge_results/codec_stream/20260909-163326/client_0.json
@@ -0,0 +1,60 @@
+{
+ "codec_start": {
+  "type": "codec.start",
+  "version": 1,
+  "utterance_index": 0,
+  "sentence_index": 0,
+  "sentence_text": "Hello, welcome to the voice synthesis benchmark test.",
+  "model": "Qwen/Qwen3-TTS-12Hz-0.6B-CustomVoice",
+  "codebooks": 16,
+  "codebook_size": 2048,
+  "frame_rate_hz": 12.5,
+  "sample_rate": 24000,
+  "decoder_id": "qwen3_tts_tokenizer_v2_12hz",
+  "header_format": "<IHB"
+ },
+ "schedule": [
+  2,
+  4,
+  8,
+  16,
+  25
+ ],
+ "total_frames": 52,
+ "ttfa_ms": 17572.953226976097,
+ "playback_start_ms": 17572.95406376943,
+ "stall_at_ttfa_ms": 0,
+ "audio_s": 4.16,
+ "total_s": 18.025338406674564,
+ "rtf": 4.333014040066001,
+ "decode_ms_total": 1063.6401386000216,
+ "chunks": [
+  {
+   "idx": 0,
+   "t_ms": 17572.95406376943,
+   "arrival_ms": 16937.01417790726,
+   "frames": 25,
+   "samples": 48000,
+   "decode_ms": 635.8697768300772
+  },
+  {
+   "idx": 1,
+   "t_ms": 17849.87504594028,
+   "arrival_ms": 16937.01417790726,
+   "frames": 25,
+   "samples": 48000,
+   "decode_ms": 276.8944939598441
+  },
+  {
+   "idx": 2,
+   "t_ms": 18000.762850046158,
+   "arrival_ms": 16937.01417790726,
+   "frames": 2,
+   "samples": 3840,
+   "decode_ms": 150.87586781010032
+  }
+ ],
+ "device": "cpu",
+ "wav": "/data/zhoutaichang/embedding_infer/vllm-omni/benchmarks/tts/edge_results/codec_stream/20260909-163326/client_0.wav",
+ "codes": "/data/zhoutaichang/embedding_infer/vllm-omni/benchmarks/tts/edge_results/codec_stream/20260909-163326/client_0.codes.npy"
+}
\ No newline at end of file
diff --git a/benchmarks/tts/edge_results/codec_stream/20260909-163326/client_1.json b/benchmarks/tts/edge_results/codec_stream/20260909-163326/client_1.json
new file mode 100644
index 000000000..a24c67345
--- /dev/null
+++ b/benchmarks/tts/edge_results/codec_stream/20260909-163326/client_1.json
@@ -0,0 +1,60 @@
+{
+ "codec_start": {
+  "type": "codec.start",
+  "version": 1,
+  "utterance_index": 0,
+  "sentence_index": 0,
+  "sentence_text": "She said she would be here by noon, but nobody showed up.",
+  "model": "Qwen/Qwen3-TTS-12Hz-0.6B-CustomVoice",
+  "codebooks": 16,
+  "codebook_size": 2048,
+  "frame_rate_hz": 12.5,
+  "sample_rate": 24000,
+  "decoder_id": "qwen3_tts_tokenizer_v2_12hz",
+  "header_format": "<IHB"
+ },
+ "schedule": [
+  2,
+  4,
+  8,
+  16,
+  25
+ ],
+ "total_frames": 56,
+ "ttfa_ms": 2366.1640412174165,
+ "playback_start_ms": 2366.16463214159,
+ "stall_at_ttfa_ms": 0,
+ "audio_s": 4.48,
+ "total_s": 2.864769846200943,
+ "rtf": 0.6394575549555676,
+ "decode_ms_total": 686.854031868279,
+ "chunks": [
+  {
+   "idx": 0,
+   "t_ms": 2366.16463214159,
+   "arrival_ms": 2176.4274998568,
+   "frames": 25,
+   "samples": 48000,
+   "decode_ms": 189.63613593950868
+  },
+  {
+   "idx": 1,
+   "t_ms": 2630.5868341587484,
+   "arrival_ms": 2176.4274998568,
+   "frames": 25,
+   "samples": 48000,
+   "decode_ms": 264.40681982785463
+  },
+  {
+   "idx": 2,
+   "t_ms": 2863.4107299149036,
+   "arrival_ms": 2176.4274998568,
+   "frames": 6,
+   "samples": 11520,
+   "decode_ms": 232.81107610091567
+  }
+ ],
+ "device": "cpu",
+ "wav": "/data/zhoutaichang/embedding_infer/vllm-omni/benchmarks/tts/edge_results/codec_stream/20260909-163326/client_1.wav",
+ "codes": "/data/zhoutaichang/embedding_infer/vllm-omni/benchmarks/tts/edge_results/codec_stream/20260909-163326/client_1.codes.npy"
+}
\ No newline at end of file
diff --git a/benchmarks/tts/edge_results/codec_stream/20260909-163326/client_2.json b/benchmarks/tts/edge_results/codec_stream/20260909-163326/client_2.json
new file mode 100644
index 000000000..e656c46bc
--- /dev/null
+++ b/benchmarks/tts/edge_results/codec_stream/20260909-163326/client_2.json
@@ -0,0 +1,60 @@
+{
+ "codec_start": {
+  "type": "codec.start",
+  "version": 1,
+  "utterance_index": 0,
+  "sentence_index": 0,
+  "sentence_text": "The quick brown fox jumps over the lazy dog near the riverbank.",
+  "model": "Qwen/Qwen3-TTS-12Hz-0.6B-CustomVoice",
+  "codebooks": 16,
+  "codebook_size": 2048,
+  "frame_rate_hz": 12.5,
+  "sample_rate": 24000,
+  "decoder_id": "qwen3_tts_tokenizer_v2_12hz",
+  "header_format": "<IHB"
+ },
+ "schedule": [
+  2,
+  4,
+  8,
+  16,
+  25
+ ],
+ "total_frames": 57,
+ "ttfa_ms": 2354.5054919086397,
+ "playback_start_ms": 2354.5059328898787,
+ "stall_at_ttfa_ms": 0,
+ "audio_s": 4.56,
+ "total_s": 3.019472629763186,
+ "rtf": 0.6621650503866636,
+ "decode_ms_total": 828.5680301487446,
+ "chunks": [
+  {
+   "idx": 0,
+   "t_ms": 2354.5059328898787,
+   "arrival_ms": 2189.843008760363,
+   "frames": 25,
+   "samples": 48000,
+   "decode_ms": 164.5867652259767
+  },
+  {
+   "idx": 1,
+   "t_ms": 2539.5107567310333,
+   "arrival_ms": 2189.843008760363,
+   "frames": 25,
+   "samples": 48000,
+   "decode_ms": 184.9919487722218
+  },
+  {
+   "idx": 2,
+   "t_ms": 3018.5127947479486,
+   "arrival_ms": 2189.843008760363,
+   "frames": 7,
+   "samples": 13440,
+   "decode_ms": 478.9893161505461
+  }
+ ],
+ "device": "cpu",
+ "wav": "/data/zhoutaichang/embedding_infer/vllm-omni/benchmarks/tts/edge_results/codec_stream/20260909-163326/client_2.wav",
+ "codes": "/data/zhoutaichang/embedding_infer/vllm-omni/benchmarks/tts/edge_results/codec_stream/20260909-163326/client_2.codes.npy"
+}
\ No newline at end of file
diff --git a/benchmarks/tts/edge_results/codec_stream/20260909-163326/client_3.json b/benchmarks/tts/edge_results/codec_stream/20260909-163326/client_3.json
new file mode 100644
index 000000000..b285af4a6
--- /dev/null
+++ b/benchmarks/tts/edge_results/codec_stream/20260909-163326/client_3.json
@@ -0,0 +1,60 @@
+{
+ "codec_start": {
+  "type": "codec.start",
+  "version": 1,
+  "utterance_index": 0,
+  "sentence_index": 0,
+  "sentence_text": "I can't believe how beautiful the sunset looks from up here on the mountain.",
+  "model": "Qwen/Qwen3-TTS-12Hz-0.6B-CustomVoice",
+  "codebooks": 16,
+  "codebook_size": 2048,
+  "frame_rate_hz": 12.5,
+  "sample_rate": 24000,
+  "decoder_id": "qwen3_tts_tokenizer_v2_12hz",
+  "header_format": "<IHB"
+ },
+ "schedule": [
+  2,
+  4,
+  8,
+  16,
+  25
+ ],
+ "total_frames": 67,
+ "ttfa_ms": 2623.2912791892886,
+ "playback_start_ms": 2623.2917830348015,
+ "stall_at_ttfa_ms": 0,
+ "audio_s": 5.36,
+ "total_s": 3.1137252133339643,
+ "rtf": 0.5809188830846949,
+ "decode_ms_total": 642.5018617883325,
+ "chunks": [
+  {
+   "idx": 0,
+   "t_ms": 2623.2917830348015,
+   "arrival_ms": 2470.173351932317,
+   "frames": 25,
+   "samples": 48000,
+   "decode_ms": 153.0545693822205
+  },
+  {
+   "idx": 1,
+   "t_ms": 2832.9488299787045,
+   "arrival_ms": 2470.173351932317,
+   "frames": 25,
+   "samples": 48000,
+   "decode_ms": 209.64194368571043
+  },
+  {
+   "idx": 2,
+   "t_ms": 3112.7667292021215,
+   "arrival_ms": 2470.173351932317,
+   "frames": 17,
+   "samples": 32640,
+   "decode_ms": 279.8053487204015
+  }
+ ],
+ "device": "cpu",
+ "wav": "/data/zhoutaichang/embedding_infer/vllm-omni/benchmarks/tts/edge_results/codec_stream/20260909-163326/client_3.wav",
+ "codes": "/data/zhoutaichang/embedding_infer/vllm-omni/benchmarks/tts/edge_results/codec_stream/20260909-163326/client_3.codes.npy"
+}
\ No newline at end of file
diff --git a/benchmarks/tts/edge_results/codec_stream/20260909-163326/summary.json b/benchmarks/tts/edge_results/codec_stream/20260909-163326/summary.json
new file mode 100644
index 000000000..6bddcfa65
--- /dev/null
+++ b/benchmarks/tts/edge_results/codec_stream/20260909-163326/summary.json
@@ -0,0 +1,71 @@
+{
+ "model": "Qwen/Qwen3-TTS-12Hz-0.6B-CustomVoice",
+ "timestamp": "20260909-163326",
+ "prompts": [
+  "Hello, welcome to the voice synthesis benchmark test.",
+  "She said she would be here by noon, but nobody showed up.",
+  "The quick brown fox jumps over the lazy dog near the riverbank.",
+  "I can't believe how beautiful the sunset looks from up here on the mountain."
+ ],
+ "client_device": "cpu",
+ "rows": [
+  {
+   "i": 0,
+   "text": "Hello, welcome to the voice synthesis benchmark test.",
+   "frames": 52,
+   "client_ttfa_ms": 17572.953226976097,
+   "client_playback_start_ms": 17572.95406376943,
+   "client_rtf": 4.333014040066001,
+   "client_decode_ms": 1063.6401386000216,
+   "client_audio_s": 4.16,
+   "full_server_s": 22.814023814164102,
+   "full_audio_s": 4.16,
+   "snr_db": -3.03693684919166,
+   "length_ratio": 1.0
+  },
+  {
+   "i": 1,
+   "text": "She said she would be here by noon, but nobody showed up.",
+   "frames": 56,
+   "client_ttfa_ms": 2366.1640412174165,
+   "client_playback_start_ms": 2366.16463214159,
+   "client_rtf": 0.6394575549555676,
+   "client_decode_ms": 686.854031868279,
+   "client_audio_s": 4.48,
+   "full_server_s": 2.6364654297940433,
+   "full_audio_s": 4.48,
+   "snr_db": -2.953329833875654,
+   "length_ratio": 1.0
+  },
+  {
+   "i": 2,
+   "text": "The quick brown fox jumps over the lazy dog near the riverbank.",
+   "frames": 57,
+   "client_ttfa_ms": 2354.5054919086397,
+   "client_playback_start_ms": 2354.5059328898787,
+   "client_rtf": 0.6621650503866636,
+   "client_decode_ms": 828.5680301487446,
+   "client_audio_s": 4.56,
+   "full_server_s": 2.656285943929106,
+   "full_audio_s": 4.56,
+   "snr_db": -3.011027358981732,
+   "length_ratio": 1.0
+  },
+  {
+   "i": 3,
+   "text": "I can't believe how beautiful the sunset looks from up here on the mountain.",
+   "frames": 67,
+   "client_ttfa_ms": 2623.2912791892886,
+   "client_playback_start_ms": 2623.2917830348015,
+   "client_rtf": 0.5809188830846949,
+   "client_decode_ms": 642.5018617883325,
+   "client_audio_s": 5.36,
+   "full_server_s": 2.823788551148027,
+   "full_audio_s": 5.36,
+   "snr_db": -2.9195661162834483,
+   "length_ratio": 1.0
+  }
+ ],
+ "talker_server_start_s": 103.04911469155923,
+ "full_server_start_s": 168.08003386110067
+}
\ No newline at end of file