p280427/acoustic_model
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

resolve the conflicts.

Browse Source
This commit is contained in:
yemaozi88
2018-12-30 23:33:45 +01:00
parent df046ffc26 beff33fdf9
commit a77ed9d4dd
26 changed files with 28 additions and 951 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
acoustic_model/acoustic_model.pyproj
View File

@@ -4,7 +4,7 @@
<SchemaVersion>2.0</SchemaVersion>
<ProjectGuid>4d8c8573-32f0-4a62-9e62-3ce5cc680390</ProjectGuid>
<ProjectHome>.</ProjectHome>
<StartupFile>check_novoapi.py</StartupFile>
<StartupFile>performance_check.py</StartupFile>
<SearchPath>
</SearchPath>
<WorkingDirectory>.</WorkingDirectory>
@@ -25,9 +25,6 @@
<Compile Include="acoustic_model_functions.py">
<SubType>Code</SubType>
</Compile>
<Compile Include="check_novoapi.py">
<SubType>Code</SubType>
</Compile>
<Compile Include="convert_xsampa2ipa.py">
<SubType>Code</SubType>
</Compile>
@@ -37,10 +34,7 @@
<Compile Include="fa_test.py">
<SubType>Code</SubType>
</Compile>
<Compile Include="forced_alignment_novo.py">
<SubType>Code</SubType>
</Compile>
<Compile Include="htk_vs_kaldi.py">
<Compile Include="performance_check.py">
<SubType>Code</SubType>
</Compile>
</ItemGroup>

51
acoustic_model/fa_test.py
View File

@@ -2,52 +2,15 @@ import os
import sys
os.chdir(r'C:\Users\Aki\source\repos\acoustic_model\acoustic_model')
import numpy as np
import defaultfiles as default
sys.path.append(os.path.join(default.repo_dir, 'forced_alignment'))
from forced_alignment import forced_alignment, lexicon, convert_phone_set
from forced_alignment import forced_alignment
#wav_file = r'C:\Users\Aki\source\repos\forced_alignment\notebooks\sample\10147-1464186409-1917281.wav'
#forced_alignment(
# wav_file,
# 'Australië'
# #'BUFFETCOUPON COULISSEN DOUANE'
# )
wav_file = r'C:\Users\Aki\source\repos\forced_alignment\notebooks\sample\10147-1464186409-1917281.wav'
forced_alignment(
wav_file,
#'Australië'
'BUFFETCOUPON COULISSEN DOUANE'
)
# according to: http://lands.let.ru.nl/cgn/doc_Dutch/topics/version_1.0/annot/phonetics/fon_prot.pdf
phone_list_cgn = ['p', 'd', 't', 'd', 'k', 'g', # plosives
'f', 'v', 's', 'z', 'S', 'Z', 'x', 'G', 'h', # fricatives
'N', 'm', 'n', 'J', 'l', 'r', 'w', 'j', # sonorant
'I', 'E', 'A', 'O', 'Y', # short vowels
'i', 'y', 'e', '2', 'a', 'o', 'u', # long vowels
'@', # schwa
'E+', 'Y+', 'A+', # Diftongen
'E:', 'Y:', 'O:', # Leenvocalen
'E~', 'A~', 'O~', 'Y~' # Nasale vocalen
]
# load word in the lexicon.
lexicon_file = r'C:\cygwin64\home\Aki\acoustic_model\material\barbara\2010_2510_lexicon_pronvars_HTK.txt'
with open(lexicon_file, 'r') as f:
lines = f.readlines()
words = []
for line in lines:
line_split = line.split()
if len(line_split) > 0:
word = line_split[0]
word.replace('+s', '')
word = word.split('-')
words.append(word)
words = list(np.unique(words))
pronunciations = lexicon._grapheme_to_phoneme(words)
htks = []
phone_list = set()
for word in pronunciations.keys():
ipa = pronunciations[word]
htk = convert_phone_set.split_ipa(ipa)
htks.append(htk)
phone_list = phone_list | set(htk)

133
acoustic_model/forced_alignment_novo.py
View File

@@ -1,133 +0,0 @@
#
# forced alignment using novo-api.
#
# *** IMPORTANT ***
# This file should be treated as confidencial.
# This file should not be copied or uploaded to public sites.
#
# NOTES:
# The usage of novo api: https://bitbucket.org/novolanguage/python-novo-api
# I couldn't make it work as I described in the mail to Martijn Bartelds on 2018/12/03.
# As per the advice from him, I modified testgrammer.py and made it a function.
#
# In order to run on Python 3.6, the following points are changed in novo-api.
# (1) backend/__init__.py
# - #import session
# from . import session
# (2) backend/session.py
# - #except Exception, e:
# except Exception as e:
# - #print self.last_message
# print(self.last_message)
# (3) asr/segment/praat.py
# - def print_tier(output, title, begin, end, segs, (format, formatter))
# def print_tier(output, title, begin, end, segs, format, formatter):
# (4) asr/spraaklab/__init.py
# - #import session
# from . import session
# (5) asr/spraaklab/schema.py
# - #print data, "validated not OK", e.message
# print("{0} validated not OK {1}".format(data, e.message))
# - #print data, "validated OK"
# print("{} validated OK".format(data))
# - #if isinstance(object, basestring):
# if isinstance(object, str)
#
# Aki Kunikoshi
# 428968@gmail.com
#
import argparse
import json
from novoapi.backend import session
# username / password cannot be passed as artuments...
p = argparse.ArgumentParser()
#p.add_argument("--user", default=None)
#p.add_argument("--password", default=None)
p.add_argument("--user", default='martijn.wieling')
p.add_argument("--password", default='fa0Thaic')
args = p.parse_args()
wav_file = 'c:\\OneDrive\\WSL\\test\\onetwothree.wav'
rec = session.Recognizer(grammar_version="1.0", lang="nl", snodeid=101, user=args.user, password=args.password, keepopen=True) # , modeldir=modeldir)
grammar = {
"type": "confusion_network",
"version": "1.0",
"data": {
"kind": "sequence",
"elements": [
{
"kind": "word",
"pronunciation": [
{
"phones": [
"wv",
"a1",
"n"
],
"id": 0
},
{
"phones": [
"wv",
"uh1",
"n"
],
"id": 1
}
],
"label": "one"
},
{
"kind": "word",
"pronunciation": [
{
"phones": [
"t",
"uw1"
],
"id": 0
}
],
"label": "two"
},
{
"kind": "word",
"pronunciation": [
{
"phones": [
"t",
"r",
"iy1"
],
"id": 0
},
{
"phones": [
"s",
"r",
"iy1"
],
"id": 1
}
],
"label": "three"
}
]
},
"return_objects": [
"grammar"
],
"phoneset": "novo70"
}
res = rec.setgrammar(grammar)
#print "Set grammar result", res
#res = rec.recognize_wav("test/onetwothree.wav")
res = rec.recognize_wav(wav_file)
#print "Recognition result:", json.dumps(res.export(), indent=4)

32
acoustic_model/htk_vs_kaldi.py → acoustic_model/performance_check.py
View File

@@ -3,7 +3,7 @@ os.chdir(r'C:\Users\Aki\source\repos\acoustic_model\acoustic_model')
import sys
import csv
#import subprocess
import subprocess
from collections import Counter
import re
@@ -20,6 +20,8 @@ from forced_alignment import pyhtk
## ======================= user define =======================
excel_file = os.path.join(default.experiments_dir, 'stimmen', 'data', 'Frisian Variants Picture Task Stimmen.xlsx')
data_dir = os.path.join(default.experiments_dir, 'stimmen', 'data')
wav_dir = r'c:\OneDrive\WSL\kaldi-trunk\egs\fame\s5\corpus\stimmen' # 16k
@@ -46,6 +48,8 @@ load_forced_alignment_kaldi = 1
eval_forced_alignment_kaldi = 1
## ======================= add paths =======================
sys.path.append(os.path.join(default.repo_dir, 'forced_alignment'))
from forced_alignment import convert_phone_set
@@ -58,15 +62,15 @@ from evaluation import plot_confusion_matrix
## ======================= convert phones ======================
mapping = convert_xsampa2ipa.load_converter('xsampa', 'ipa', default.ipa_xsampa_converter_dir)
xls = pd.ExcelFile(default.stimmen_transcription_xlsx)
xls = pd.ExcelFile(excel_file)
## check conversion
#df = pd.read_excel(xls, 'check')
#df = pd.read_excel(xls, 'frequency')
#for xsampa, ipa in zip(df['X-SAMPA'], df['IPA']):
# if xsampa is not '/':
# ipa_converted = convert_xsampa2ipa.xsampa2ipa(mapping, xsampa)
# if not ipa_converted == ipa:
# print('{0}: {1} - {2}'.format(xsampa, ipa_converted, ipa))
# #ipa_converted = convert_xsampa2ipa.conversion('xsampa', 'ipa', mapping, xsampa_)
# ipa_converted = convert_xsampa2ipa.xsampa2ipa(mapping, xsampa)
# if not ipa_converted == ipa:
# print('{0}: {1} - {2}'.format(xsampa, ipa_converted, ipa))
## check phones included in FAME!
@@ -156,7 +160,7 @@ if do_forced_alignment_htk:
htk_dict_file = os.path.join(htk_dict_dir, word + '.dic')
pyhtk.doHVite(wav_file, label_file, htk_dict_file, fa_file, default.config_hvite,
default.phonelist_friesian_txt, acoustic_model)
default.phonelist, acoustic_model)
os.remove(label_file)
prediction = am_func.read_fileFA(fa_file)
@@ -227,7 +231,7 @@ if make_kaldi_data_files:
## ======================= make lexicon txt which is used by Kaldi =======================
if make_kaldi_lexicon_txt:
option_num = 7
option_num = 6
# remove previous file.
if os.path.exists(lexicon_txt):
@@ -277,10 +281,10 @@ if load_forced_alignment_kaldi:
phones_txt = os.path.join(default.kaldi_dir, 'data', 'lang', 'phones.txt')
merged_alignment_txt = os.path.join(default.kaldi_dir, 'exp', 'tri1_alignme', 'merged_alignment.txt')
#filenames = np.load(stimmen_data_dir + '\\filenames.npy')
#words = np.load(stimmen_data_dir + '\\words.npy')
#pronunciations = np.load(stimmen_data_dir + '\\pronunciations_ipa.npy')
#pronvar_list_all = np.load(stimmen_data_dir + '\\pronvar_list_all.npy')
#filenames = np.load(data_dir + '\\filenames.npy')
#words = np.load(data_dir + '\\words.npy')
#pronunciations = np.load(data_dir + '\\pronunciations_ipa.npy')
#pronvar_list_all = np.load(data_dir + '\\pronvar_list_all.npy')
#word_list = np.unique(words)
# load the mapping between phones and ids.
@@ -365,7 +369,7 @@ if eval_forced_alignment_htk:
if compare_hmm_num:
f_result.write("{},".format(hmm_num_str))
#match = np.load(stimmen_data_dir + '\\match_hmm' + hmm_num_str + '.npy')
#match = np.load(data_dir + '\\match_hmm' + hmm_num_str + '.npy')
#prediction = np.load(os.path.join(result_dir, 'htk', 'predictions_hmm' + hmm_num_str + '.npy'))
#prediction = pd.Series(prediction, index=df.index, name='prediction')
#result = pd.concat([df, prediction], axis=1)