我试图使用HTK并相应地设置项目。我成功地使用HCopy对训练数据进行了编码,但在尝试执行时:
HCompV -C config -f 0.01 -m -S codetr.scp -M hmm0 proto
我收到以下错误:
ERROR [+6251] Input file is not in RIFF format
ERROR [+6213] OpenWaveInput: Get[format]HeaderInfo failed
ERROR [+6313] OpenAsChannel: OpenWaveInput failed
ERROR [+6316] OpenBuffer: OpenAsChannel failed
ERROR [+2050] LoadFile: Config parameters invalid
FATAL ERROR - Terminating program
作为参考,我在MSYS shell中运行它,配置如下:
# Coding parameters
SOURCEFORMAT = WAV # We read in .wav files
TARGETKIND = MFCC_0_D_A # Identifier for coefficients
TARGETRATE = 100000.0 # 10ms = frame periodicity
SAVECOMPRESSED = T
SAVEWITHCRC = T
WINDOWSIZE = 250000.0 # 25ms = time frame length
USEHAMMING = T # Use of Hamming function for windowing
PREEMCOEF = 0.97 # Pre-emphasis coefficient
NUMCHANS = 26 # Number of filterbank channels
CEPLIFTER = 22 # Length of cepstral filtering
NUMCEPS = 12 # Number of MFCC coefficients
ENORMALISE = F # Normalise intensity of data
proto文件如下:
~o <VecSize> 39 <MFCC_0_D_A>
~h "proto"
<BeginHMM>
<NumStates> 5
<State> 2
<Mean> 39
0.0 0.0 0.0 0.0 ...
<Variance> 39
1.0 1.0 1.0 1.0 ...
<State> 3
<Mean> 39
0.0 0.0 0.0 0.0 ...
<Variance> 39
1.0 1.0 1.0 1.0 ...
<State> 4
<Mean> 39
0.0 0.0 0.0 0.0 ...
<Variance> 39
1.0 1.0 1.0 1.0 ...
<TransP> 5
0.0 1.0 0.0 0.0 0.0
0.0 0.6 0.4 0.0 0.0
0.0 0.0 0.6 0.4 0.0
0.0 0.0 0.0 0.7 0.3
0.0 0.0 0.0 0.0 0.0
<EndHMM>
和codetr.scp文件如下(所有文件都存在):
samples/cough1.wav samples/cough1.mfc
samples/cough2.wav samples/cough2.mfc
samples/cough3.wav samples/cough3.mfc
samples/cough4.wav samples/cough4.mfc
samples/silence1.wav samples/silence1.mfc
samples/silence2.wav samples/silence2.mfc
samples/silence3.wav samples/silence3.mfc
samples/silence4.wav samples/silence4.mfc
感谢任何和所有帮助!
答案 0 :(得分:0)
您检查过输入文件格式吗?在unix / linux上,你可以尝试文件[filename]。发布测试结果。可以使用诸如audacity之类的工具转换为RIFF格式。