rozen-test is added.

acoustic_model/acoustic_model.pyproj

@@ -36,6 +36,9 @@
     <Compile Include="fa_test.py">
       <SubType>Code</SubType>
     </Compile>
+    <Compile Include="forced_aligner_comparison.py">
+      <SubType>Code</SubType>
+    </Compile>
     <Compile Include="novoapi_forced_alignment.py">
       <SubType>Code</SubType>
     </Compile>

acoustic_model/check_novoapi.py

@@ -10,14 +10,13 @@ import shutil
 import numpy as np
 import pandas as pd
 import matplotlib.pyplot as plt
-
 from sklearn.metrics import confusion_matrix
 from sklearn.metrics import accuracy_score
 import novoapi 
 
 import defaultfiles as default
 sys.path.append(default.forced_alignment_module_dir)
-from forced_alignment import pyhtk, convert_phone_set
+from forced_alignment import convert_phone_set
 #import acoustic_model_functions as am_func
 import convert_xsampa2ipa
 import novoapi_functions
@@ -47,10 +46,6 @@ david_suggestion = ['ɔː', 'ɪː', 'iː', 'œː', 'ɛː', 'w']
 ## read pronunciation variants.
 stimmen_transcription_ = pd.ExcelFile(default.stimmen_transcription_xlsx)
 df = pd.read_excel(stimmen_transcription_, 'frequency')
-#for xsampa, ipa in zip(df['X-SAMPA'], df['IPA']):
-#    ipa_converted = convert_xsampa2ipa.xsampa2ipa(mapping, xsampa)
-#    if not ipa_converted == ipa:
-#        print('{0}: {1} - {2}'.format(xsampa, ipa_converted, ipa))
 transcription_ipa = list(df['IPA'])
 
 # transcription mistake?
@@ -63,6 +58,7 @@ for ipa in transcription_ipa:
 	ipa = ipa.replace(':', 'ː')
 	ipa = convert_phone_set.split_ipa(ipa)
 
+	# list of phones not in novo70 phoneset.
 	not_in_novo70_ = [phone for phone in ipa 
 				   if not phone in phoneset_ipa and not phone in david_suggestion]
 	not_in_novo70_ = [phone.replace('sp', '') for phone in not_in_novo70_]
@@ -106,6 +102,10 @@ df = pd.read_excel(stimmen_transcription_, 'original')
 
 # mapping from ipa to xsampa
 mapping = convert_xsampa2ipa.load_converter('xsampa', 'ipa', default.ipa_xsampa_converter_dir)
+#for xsampa, ipa in zip(df['X-SAMPA'], df['IPA']):
+#    ipa_converted = convert_xsampa2ipa.xsampa2ipa(mapping, xsampa)
+#    if not ipa_converted == ipa:
+#        print('{0}: {1} - {2}'.format(xsampa, ipa_converted, ipa))
 
 ipas     = []
 famehtks = []
@@ -153,12 +153,12 @@ for word in word_list:
 
 
 ## ===== forced alignment =====
-reus_dir = r'C:\OneDrive\Desktop\Reus'
+rozen_dir = r'c:\Users\Aki\source\repos\acoustic_model\rozen-test'
 if forced_alignment_novo70:
 	Results = pd.DataFrame(index=[],
 		columns=['filename', 'word', 'xsampa', 'ipa', 'result_ipa', 'result_novo70', 'llh'])
 	#for word in word_list:
-	for word in ['Reus']:
+	for word in ['Rozen']:
 		# pronunciation variants top 3
 		df_per_word_ = df_per_word[df_per_word['word']==word]
 		df_per_word_ = df_per_word_.sort_values('frequency', ascending=False)
@@ -208,37 +208,35 @@ if forced_alignment_novo70:
 			wav_file = os.path.join(default.stimmen_wav_dir, filename)
 			if os.path.exists(wav_file):
 				# for Martijn
-				#shutil.copy(wav_file, os.path.join(reus_dir, filename))
+				shutil.copy(wav_file, os.path.join(rozen_dir, filename))
 
-				pronunciation_ipa_ = [ipa.replace(':', 'ː') for ipa in pronunciation_ipa]
+		#		pronunciation_ipa_ = [ipa.replace(':', 'ː') for ipa in pronunciation_ipa]
-				result = novoapi_functions.forced_alignment(wav_file, word, pronunciation_ipa_)
+		#		result = novoapi_functions.forced_alignment(wav_file, word, pronunciation_ipa_)
-				result_ipa, result_novo70, llh = novoapi_functions.result2pronunciation(result, word)
+		#		result_ipa, result_novo70, llh = novoapi_functions.result2pronunciation(result, word)
-				result_ = pd.Series([
+		#		result_ = pd.Series([
-					sample['filename'],
+		#			sample['filename'],
-					sample['word'],
+		#			sample['word'],
-					sample['xsampa'],
+		#			sample['xsampa'],
-					sample['ipa'],
+		#			sample['ipa'],
-					' '.join(result_ipa),
+		#			' '.join(result_ipa),
-					' '.join(result_novo70),
+		#			' '.join(result_novo70),
-					llh
+		#			llh
-					], index=results.columns)
+		#			], index=results.columns)
-				results = results.append(result_, ignore_index = True)
+		#		results = results.append(result_, ignore_index = True)
-				print('{0}/{1}: answer {2} - prediction {3}'.format( 
+		#		print('{0}/{1}: answer {2} - prediction {3}'.format( 
-			 i+1, len(samples), result_['ipa'], result_['result_ipa']))
+		#	 i+1, len(samples), result_['ipa'], result_['result_ipa']))
-			results.to_excel(os.path.join(reus_dir, 'results.xlsx'), encoding="utf-8")
+		#	#results.to_excel(os.path.join(default.stimmen_dir, 'results.xlsx'), encoding="utf-8")
-		if len(results) > 0:
+		#if len(results) > 0:
-			Results = Results.append(results, ignore_index = True)
+		#	Results = Results.append(results, ignore_index = True)
-		Results.to_excel(os.path.join(default.stimmen_dir, 'Results.xlsx'), encoding="utf-8")
+		#Results.to_excel(os.path.join(default.stimmen_result_novoapi_dir, 'Results.xlsx'), encoding="utf-8")
 else:
-	Results_xlsx = pd.ExcelFile(os.path.join(default.stimmen_dir, 'Results.xlsx'), encoding="utf-8")
+	Results_xlsx = pd.ExcelFile(os.path.join(default.stimmen_result_novoapi_dir, 'Results.xlsx'), encoding="utf-8")
 	Results = pd.read_excel(Results_xlsx, 'Sheet1')
 
 
 ## ===== analysis =====
-#result_novoapi_dir = os.path.join(default.stimmen_dir, 'result', 'novoapi')
 #for word in word_list:
 #	if not word == 'Oog':
-
 #		Results_ = Results[Results['word'] == word]
 #		y_true  = list(Results_['ipa'])
 #		y_pred_ = [ipa.replace(' ', '') for ipa in list(Results_['result_ipa'])]
@@ -249,4 +247,4 @@ else:
 #		plt.figure()
 #		output_confusion_matrix.plot_confusion_matrix(cm, pronunciation_variants, normalize=False)
 #		#plt.show()
-#		plt.savefig(os.path.join(result_novoapi_dir, word + '.png'))
+#		plt.savefig(os.path.join(default.stimmen_result_novoapi_dir, word + '.png'))

acoustic_model/defaultfiles.py

@@ -43,6 +43,7 @@ stimmen_data_dir = os.path.join(stimmen_dir, 'data')
 #stimmen_wav_dir  = os.path.join(stimmen_dir, 'wav')
 # 16 kHz
 stimmen_wav_dir  = r'c:\OneDrive\WSL\kaldi-trunk\egs\fame\s5\corpus\stimmen'
+stimmen_result_novoapi_dir = os.path.join(stimmen_dir, 'result', 'novoapi')
 
 stimmen_transcription_xlsx = os.path.join(stimmen_data_dir, 'Frisian Variants Picture Task Stimmen.xlsx')
 phonelist_friesian_txt     = os.path.join(experiments_dir, 'friesian', 'acoustic_model', 'config', 'phonelist_friesian.txt')

acoustic_model/forced_aligner_comparison.py

@@ -0,0 +1,10 @@
+import os
+os.chdir(r'C:\Users\Aki\source\repos\acoustic_model\acoustic_model')
+
+import defaultfiles as default
+
+wav_dir   = r'c:\Users\Aki\source\repos\acoustic_model\reus-test'
+wav_files = ['reus1008-reus.wav', 
+			 'reus1167-man.wav',
+			 'reus3768-mantsje.wav']
+

acoustic_model/novoapi_functions.py

@@ -7,7 +7,7 @@ import json
 from novoapi.backend import session
 
 import os
-os.chdir(r'C:\Users\Aki\source\repos\acoustic_model\acoustic_model')
+#os.chdir(r'C:\Users\Aki\source\repos\acoustic_model\acoustic_model')
 import defaultfiles as default
 
 

rozen-test/rozen-test.py

@@ -0,0 +1,119 @@
+#!/usr/bin/env python
+import os
+os.chdir(r'C:\Users\Aki\source\repos\acoustic_model\acoustic_model')
+
+import argparse
+import json
+
+from novoapi.backend import session
+
+p = argparse.ArgumentParser()
+p.add_argument("--user", default='martijn.wieling')
+p.add_argument("--password", default='xxxxx')
+args = p.parse_args()
+
+rec = session.Recognizer(grammar_version="1.0", lang="nl", snodeid=101, user=args.user, password=args.password, keepopen=True)
+
+grammar = {
+  "type": "confusion_network",
+  "version": "1.0",
+  "data": {
+    "kind": "sequence",
+    "elements": [
+      {
+        "kind": "word",
+        "pronunciation": [
+          {
+            "phones": [
+              "r",
+              "eu0",
+              "s"
+            ],
+            "id": 0
+          }
+          ,
+          {
+            "phones": [
+              "m",
+              "a0",
+              "n"
+            ],
+            "id": 1
+          }
+		   ,
+          {
+            "phones": [
+              "m",
+              "a0",
+              "n",
+              "t",
+              "s",
+              "y",
+              "ax"
+            ],
+            "id": 2
+          }
+        ],
+        "label": "reus"
+      }
+    ]
+  },
+  "return_objects": [
+    "grammar"
+  ],
+  "phoneset": "novo70"
+}
+
+res = rec.setgrammar(grammar)
+#print "Set grammar result", res
+
+
+## === novoapi/backend/session.py ===
+#import wave
+#import time
+#from novoapi.backend.session import rpcid, segmentation
+
+#wavf = "reus1008-reus.wav"
+#w = wave.open(wavf, 'r')
+#nchannels, sampwidth, framerate, nframes, comptype, compname = w.getparams()
+#buf = w.readframes(nframes)
+#w.close()
+
+#buffer_size = 4096
+#nbytes_sent = 0
+#start = time.time()
+#for j in range(0, len(buf), buffer_size):
+#    audio_packet = buf[j:j + buffer_size]
+#    nbytes_sent += len(audio_packet)
+#    rec.conn.send_binary(audio_packet)
+#rec.conn.send(json.dumps({"jsonrpc": "2.0", "method": "get_result", "id": rpcid.next()}))
+#print(rpcid.next())
+#rec.last_message = rec.conn.recv() 
+#message = json.loads(rec.last_message)
+#result = session.segmentation(message["result"]["words"])
+#result.export()
+## ====================================
+
+def result2pronunciation(result, word):
+	#result_ = res.export()[1]
+	result_ = [result[i] for i in range(len(result)) if result[i]['label'] == word] 
+	llh = result_[0]['llh']
+	phones = result_[0]['phones']
+	pronunciation = [phone['label'] for phone in phones]
+	return pronunciation, llh
+
+
+res = rec.recognize_wav("reus1008-reus.wav")
+#print "\n\n\nThe pronounced word in reus1008-reus.wav is: REUS\n\n"
+#print "Recognition result:", json.dumps(res.export(), indent=4)
+result2pronunciation(res.export(), 'reus')
+
+#print "\n\n\nThe pronounced word in reus1167-man.wav is: MAN\n\n"
+res2 = rec.recognize_wav("reus1167-man.wav")
+#print "Recognition result:", json.dumps(res2.export(), indent=4)
+result2pronunciation(res2.export(), 'reus')
+
+#print "\n\n\nThe pronounced word in reus3768-mantsje.wav is: MANTSJE\n\n"
+res3 = rec.recognize_wav("reus3768-mantsje.wav")
+#print "Recognition result:", json.dumps(res3.export(), indent=4)
+result2pronunciation(res3.export(), 'reus')