# Inference Log
# Repository: parakeet-tdt-0.6b-v2-ascend
# Date: 2026-05-20

Command: python inference.py --model_path ./model_weights --device npu

Result: SUCCESS

Input: test_audio_real_speech.wav (3.975s, 16kHz, real English speech from Kokoro TTS)
CPU transcription: "Hello, this is a test of the Kakarotex speech system."
NPU transcription: "Hello, this is a test of the Kakarotex speech system."
Text exact match: True
WER (CPU vs NPU): 0.000000%
CER (CPU vs NPU): 0.000000%
CPU inference time: 3.064s
NPU inference time: 4.092s
NPU memory: 2397.94 MB allocated

{
  "model": "parakeet-tdt-0.6b-v2",
  "audio_file": "./test_audio_real_speech.wav",
  "audio_duration_s": 3.975,
  "cpu_transcription": "Hello, this is a test of the Kakarotex speech system.",
  "npu_transcription": "Hello, this is a test of the Kakarotex speech system.",
  "cpu_inference_time_s": 3.064,
  "npu_inference_time_s": 4.0915,
  "text_exact_match": true,
  "wer_cpu_vs_npu": 0.0,
  "cer_cpu_vs_npu": 0.0,
  "error_percentage": 0.0,
  "threshold_percent": 1.0,
  "passed": true,
  "npu_device": "Ascend910_9362",
  "npu_memory_allocated_mb": 2397.94,
  "npu_memory_reserved_mb": 2830.0
}

.
├── .gitignore
├── README.md
├── assets/accuracy_eval_result.png
├── assets/env_check.png
├── assets/git_submit_result.png
├── assets/inference_result.png
├── assets/performance_eval_result.png
├── eval/eval_accuracy.py
├── eval/eval_accuracy_standalone.py
├── eval/eval_performance.py
├── inference.py
├── logs/accuracy_eval.log
├── logs/env_check.log
├── logs/inference.log
├── logs/performance_eval.log
├── requirements.txt
├── results/accuracy_eval.json
├── results/env_info.json
└── results/performance_eval.json

mkdir -p weights
# 将下载后的模型权重或模型目录放入 weights/<model_name>，运行时通过 --model_path 传入

pip install -r requirements.txt
python inference.py --model_path <model_path> --audio <audio.wav> --device npu

python eval/eval_accuracy.py --model_path <model_path> --device npu
python eval/eval_performance.py --model_path <model_path> --device npu

python eval/eval_accuracy.py --model_path <model_path> --device npu
python eval/eval_performance.py --model_path <model_path> --device npu

python inference.py --help
python inference.py --model_path <model_path> --audio <audio.wav> --device npu

# Accuracy Evaluation Evidence

Repository: parakeet-tdt-0.6b-v2-ascend
Model: parakeet-tdt-0.6b-v2
Date: 2026-05-20

Command:
python eval/eval_accuracy.py --model_path ./model_weights --device npu --output_json results/accuracy_eval.json

Real Accuracy Results (from results/accuracy_eval.json):
Audio: test_audio_real_speech.wav (3.975s of real English speech)
CPU transcription: "Hello, this is a test of the Kakarotex speech system."
NPU transcription: "Hello, this is a test of the Kakarotex speech system."

Text exact match: True
WER (CPU vs NPU): 0.000000 (0.0000%)
CER (CPU vs NPU): 0.000000 (0.0000%)
Error percentage: 0.0%
Threshold: < 1%
Result: PASSED

Word timestamps (CPU):
- "Hello," (0.24s - 0.72s)
- "this" (0.88s - 1.04s)
- "is" (1.04s - 1.12s)
- "a" (1.12s - 1.28s)
- "test" (1.28s - 1.44s)
- "of" (1.44s - 1.60s)
- "the" (1.60s - 1.76s)
- "Kakarotex" (1.76s - 2.64s)
- "speech" (2.96s - 3.36s)
- "system." (3.36s - 3.60s)

Word timestamps (NPU): identical to CPU

Status: SUCCESS

# Environment Check Evidence

Repository: parakeet-tdt-0.6b-v2-ascend
Model: 主要特性：
Date: 2026-05-16 07:03:22

Command:
npu-smi info
python3 -c "import torch; print(torch.__version__)"
python3 -c "import torch_npu; print(torch_npu.__version__)"

Key Output:
OS: Linux pod-8e032c81b34d489191e775768926f3b6 5.10.0-182.0.0.95.r2220_156.hce2.aarch64 #1 SMP Sat Sep 14 02:34:54 UTC 2024 aarch64 aarch64 aarch64 GNU/Linux
Python: 3.11.14
NPU: Ascend910 x2 (npu-smi info confirms OK)
CANN: 8.5.1
torch: 2.9.0+cpu
torch_npu: 2.9.0.post1+gitee7ba04
transformers: 4.57.6
Git Branch: main
Git Commit: 33b376f3664c21b754b896d817df0e8dccfbb81f

Status:
SUCCESS

Note:
NPU hardware detected and healthy. torch_npu importable.

# Git Submit Evidence

Repository:
https://atomgit.com/nanyizjm/parakeet-tdt-0.6b-v2-ascend.git

Branch:
main

Commit:
83de03dc7abb7a29d6a7805c28fe69b269d3b131

Command:
git status
git add .
git commit -m "docs: complete track1 delivery evidence"
git push

Status:
SUCCESS

Note:
All delivery materials committed and pushed.

# Inference Evidence

Repository: parakeet-tdt-0.6b-v2-ascend
Model: parakeet-tdt-0.6b-v2
Date: 2026-05-20

Command:
python inference.py --model_path ./model_weights --device npu

Real Inference Output (with real speech audio "Hello, this is a test of the Kokoro text to speech system."):
Input: test_audio_real_speech.wav (3.975s, 16kHz, resampled from Kokoro TTS output)
CPU transcription: "Hello, this is a test of the Kakarotex speech system."
NPU transcription: "Hello, this is a test of the Kakarotex speech system."
Text exact match: True
WER (CPU vs NPU): 0.000000%
CER (CPU vs NPU): 0.000000%
CPU inference time: 3.064s
NPU inference time: 4.092s
NPU memory: 2397.94 MB allocated

Status: SUCCESS

# Performance Evaluation Evidence

Repository: parakeet-tdt-0.6b-v2-ascend
Model: parakeet-tdt-0.6b-v2
Date: 2026-05-20

Command:
python eval/eval_performance.py --model_path ./model_weights --device npu --output_json results/performance_eval.json

Real Performance Results (from results/performance_eval.json):
Audio: 5.0s test audio, 5 runs, 2 warmup runs
Mean latency: 128.40ms
Std latency: 41.18ms
Min latency: 94.75ms
Max latency: 206.37ms
Median latency: 116.03ms
Mean RTF: 0.025681
Throughput: 38.94x realtime
NPU Memory: 2397.93 MB allocated, 2846.0 MB reserved, 2445.92 MB peak
Device: Ascend NPU (Ascend910_9362)

Status: SUCCESS

# Environment Check Log
# Repository: parakeet-tdt-0.6b-v2-ascend
# Model: 主要特性：
# Date: 2026-05-16 07:03:22

## System Info
Linux pod-8e032c81b34d489191e775768926f3b6 5.10.0-182.0.0.95.r2220_156.hce2.aarch64 #1 SMP Sat Sep 14 02:34:54 UTC 2024 aarch64 aarch64 aarch64 GNU/Linux

## Python
Python 3.11.14
pip 26.0.1 from /usr/local/python3.11.14/lib/python3.11/site-packages/pip (python 3.11)

## NPU Info
+------------------------------------------------------------------------------------------------+
| npu-smi 25.5.2                   Version: 25.5.2                                               |
+---------------------------+---------------+----------------------------------------------------+
| NPU   Name                | Health        | Power(W)    Temp(C)           Hugepages-Usage(page)|
| Chip  Phy-ID              | Bus-Id        | AICore(%)   Memory-Usage(MB)  HBM-Usage(MB)        |
+===========================+===============+====================================================+
| 0     Ascend910           | OK            | 175.8       48                0    / 0             |
| 0     0                   | 0000:0A:00.0  | 0           0    / 0          3107 / 65536         |
+------------------------------------------------------------------------------------------------+
| 0     Ascend910           | OK            | -           48                0    / 0             |
| 1     1                   | 0000:0B:00.0  | 0           0    / 0          2870 / 65536         |
+===========================+===============+====================================================+
+---------------------------+---------------+----------------------------------------------------+
| NPU     Chip              | Process id    | Process name             | Process memory(MB)      |
+===========================+===============+====================================================+
| No running processes found in NPU 0                                                            |
+===========================+===============+====================================================+

## CANN Version
8.5.1

## PyTorch
2.9.0+cpu

## torch_npu
2.9.0.post1+gitee7ba04

## transformers
4.57.6

## Git Info
Branch: main
Commit: 33b376f3664c21b754b896d817df0e8dccfbb81f

<redacted sensitive line>
ASCEND_TOOLKIT_HOME=/usr/local/Ascend/cann-8.5.1
PYTHONPATH=/usr/local/Ascend/cann-8.5.1/python/site-packages:/usr/local/Ascend/cann-8.5.1/opp/built-in/op_impl/ai_core/tbe:/usr/local/Ascend/ascend-toolkit/latest/python/site-packages:/usr/local/Ascend/ascend-toolkit/latest/opp/built-in/op_impl/ai_core/tbe:

{
  "model_name": "主要特性：",
  "repo": "parakeet-tdt-0.6b-v2-ascend",
  "repo_url": "https://atomgit.com/nanyizjm/parakeet-tdt-0.6b-v2-ascend.git",
  "status": "SUCCESS",
  "os": "Linux",
  "python": "3.11.14",
  "cann_version": "8.5.1",
  "torch_version": "2.9.0+cpu",
  "torch_npu_version": "2.9.0.post1",
  "transformers_version": "4.57.6",
  "accelerate_version": "N/A",
  "npu_available": true,
  "npu_info": "Ascend910 x2",
  "git_branch": "main",
  "git_commit": "33b376f3664c21b754b896d817df0e8dccfbb81f",
  "timestamp": "2026-05-16 07:03:22",
  "note": "Environment check passed. NPU Ascend910 available."
}

# Inference Log
# Repository: parakeet-tdt-0.6b-v2-ascend
# Date: 2026-05-20

Command: python inference.py --model_path ./model_weights --device npu

Result: SUCCESS

Input: test_audio_real_speech.wav (3.975s, 16kHz, real English speech from Kokoro TTS)
CPU transcription: "Hello, this is a test of the Kakarotex speech system."
NPU transcription: "Hello, this is a test of the Kakarotex speech system."
Text exact match: True
WER (CPU vs NPU): 0.000000%
CER (CPU vs NPU): 0.000000%
CPU inference time: 3.064s
NPU inference time: 4.092s
NPU memory: 2397.94 MB allocated
NPU memory reserved: 2830.0 MB
NPU device: Ascend910_9362

# Accuracy Evaluation Log
# Repository: parakeet-tdt-0.6b-v2-ascend
# Date: 2026-05-20

Command: python eval/eval_accuracy.py --model_path ./model_weights --device npu --output_json results/accuracy_eval.json

Result: SUCCESS

Audio: test_audio_real_speech.wav (3.975s of real English speech)
CPU transcription: "Hello, this is a test of the Kakarotex speech system."
NPU transcription: "Hello, this is a test of the Kakarotex speech system."
Text exact match: True
WER (CPU vs NPU): 0.000000 (0.0000%)
CER (CPU vs NPU): 0.000000 (0.0000%)
Error percentage: 0.0%
Threshold: < 1%
Result: PASSED
NPU device: Ascend910_9362

{
  "model": "parakeet-tdt-0.6b-v2",
  "audio_file": "./test_audio_real_speech.wav",
  "audio_duration_s": 3.975,
  "cpu_transcription": "Hello, this is a test of the Kakarotex speech system.",
  "npu_transcription": "Hello, this is a test of the Kakarotex speech system.",
  "cpu_inference_time_s": 3.064,
  "npu_inference_time_s": 4.0915,
  "text_exact_match": true,
  "wer_cpu_vs_npu": 0.0,
  "cer_cpu_vs_npu": 0.0,
  "error_percentage": 0.0,
  "threshold_percent": 1.0,
  "passed": true,
  "npu_device": "Ascend910_9362",
  "npu_memory_allocated_mb": 2397.94,
  "npu_memory_reserved_mb": 2830.0
}

# Performance Evaluation Log
# Repository: parakeet-tdt-0.6b-v2-ascend
# Date: 2026-05-20

Command: python eval/eval_performance.py --model_path ./model_weights --device npu --output_json results/performance_eval.json

Result: SUCCESS

Audio: 5.0s test audio, 5 runs, 2 warmup runs
Mean latency: 128.40ms
Std latency: 41.18ms
Min latency: 94.75ms
Max latency: 206.37ms
Median latency: 116.03ms
Mean RTF: 0.025681
Throughput: 38.94x realtime
NPU Memory: 2397.93 MB allocated, 2846.0 MB reserved, 2445.92 MB peak
NPU device: Ascend910_9362

{
  "model": "parakeet-tdt-0.6b-v2",
  "device": "npu",
  "num_runs": 5,
  "warmup_runs": 2,
  "latency_stats": {
    "mean_ms": 128.4,
    "std_ms": 41.18,
    "min_ms": 94.75,
    "max_ms": 206.37,
    "median_ms": 116.03
  },
  "rtf_stats": {
    "mean": 0.025681
  },
  "throughput": {
    "seconds_of_audio_per_second": 38.94
  },
  "memory": {
    "allocated_MB": 2397.93,
    "reserved_MB": 2846.0,
    "max_allocated_MB": 2445.92
  },
  "npu_device": "Ascend910_9362"
}

python inference.py \
  --model_path ./model_weights/parakeet-tdt-0.6b-v2.nemo \
  --device npu \
  --output_log ./logs/inference.log

python eval/eval_accuracy.py \
  --model_path ./model_weights/parakeet-tdt-0.6b-v2.nemo \
  --output_log ./logs/accuracy_eval.log \
  --output_json ./results/accuracy_eval.json

python eval/eval_performance.py \
  --model_path ./model_weights/parakeet-tdt-0.6b-v2.nemo \
  --device npu \
  --dtype float32 \
  --num_runs 5 \
  --warmup_runs 2 \
  --output_log ./logs/performance_eval.log \
  --output_json ./results/performance_eval.json