HTK工具详解：从WAV到MFCC的转换-CSDN博客

本文链接：https://blog.csdn.net/xiaocao9903/article/details/78846582

本文介绍了HTK工具的使用，包括如何将WAV格式的音频文件转换为MFCC特征表示。通过配置文件config_hlist，我们可以检查输入转换是否正确。示例中展示了HList命令的用法，详细展示了转换前后的声音样本数据。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

tool example：
./HList -h -s 2000 -e 3000 -F WAV ./data/train/speech/s100.wav

------------------- Source: ./data/train/speech/s100.wav --------------------
Sample Bytes: 2 Sample Kind: WAVEFORM
Num Comps: 1 Sample Period: 125.0 us
Num Samples: 27649 File Format: WAV
--------------------------- Samples: 10000->10050 ---------------------------
10000: -397 -19 622 1420 865 -970 -1819 -1765 -2084 -2440
10010: -2825 -3014 -2899 -2545 -1752 -1426 -1659 -2261 -2961 -2974
10020: -2423 -1814 -1408 -478 499 1299 3111 3874 2624 1534
10030: 1752 2584 2457 2652 3304 3004 2536 1853 927 342
10040: 872 2284 2628 1508 -410 -2086 -1769 -103 890 979
10050: 1179
------------------------------------ END ------------------------------------

config_hlist内容如下

# Coding parameters wav-->mfcc
SOURCEFORMAT = WAV
#SOURCEFORMAT = HTK
TARGETKIND = MFCC_0_D_A
TARGETRATE = 100000.0 #10ms frame rate
#SAVECOMPRESSED = T
#SAVEWITHCRC = T
WINDOWSIZE = 250000.0 #25ms window
USEHAMMING = T
PREEMCOEF = 0.97
NUMCHANS = 26
CEPLIFTER = 22
NUMCEPS = 12
ENORMALISE = F

HList:
The second use of HList is to check that input conversions are being performed properly.

./HList -C config_hlist -o -h -t -s 100 -e 104 -i 9 ./data/train/speech/s100.wav
--------------------- Source: ./data/train/speech/s100.wav ---------------------
Sample Bytes: 2 Sample Kind: WAVEFORM
Num Comps: 1 Sample Period: 125.0 us
Num Samples: 27649 File Format: WAV
------------------------------------ Target ------------------------------------
Sample Bytes: 156 Sample Kind: &nb

htk 工具使用介绍