awacke1 commited on
Commit
bee8f8a
1 Parent(s): 545471a

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +13 -13
app.py CHANGED
@@ -24,6 +24,7 @@ DATASET_REPO_ID = "awacke1/MindfulStory.csv"
24
  DATA_FILENAME = "MindfulStory.csv"
25
  DATA_FILE = os.path.join("data", DATA_FILENAME)
26
  HF_TOKEN = os.environ.get("HF_TOKEN")
 
27
  # Download dataset repo using hub download
28
  try:
29
  hf_hub_download(
@@ -47,12 +48,12 @@ with open('Mindfulness.txt', 'r') as file:
47
  context = file.read()
48
 
49
  # Set up cloned dataset from repo for operations
50
- repo = Repository(
51
- local_dir="data", clone_from=DATASET_REPO_URL, use_auth_token=HF_TOKEN
52
- )
53
 
 
54
  asr = pipeline("automatic-speech-recognition", "facebook/wav2vec2-base-960h")
55
 
 
56
  MODEL_NAMES = [
57
  "en/ljspeech/tacotron2-DDC",
58
  "en/ljspeech/glow-tts",
@@ -62,6 +63,8 @@ MODEL_NAMES = [
62
  "fr/mai/tacotron2-DDC",
63
  "de/thorsten/tacotron2-DCA",
64
  ]
 
 
65
  MODELS = {}
66
  manager = ModelManager()
67
  for MODEL_NAME in MODEL_NAMES:
@@ -78,24 +81,23 @@ for MODEL_NAME in MODEL_NAMES:
78
  )
79
  MODELS[MODEL_NAME] = synthesizer
80
 
 
81
  def transcribe(audio):
82
  text = asr(audio)["text"]
83
  return text
84
 
 
85
  classifier = pipeline("text-classification")
86
 
 
87
  def speech_to_text(speech):
88
  text = asr(speech)["text"]
89
-
90
  #rMem = AIMemory("STT", text)
91
-
92
  return text
93
 
94
  def text_to_sentiment(text):
95
  sentiment = classifier(text)[0]["label"]
96
-
97
  #rMem = AIMemory(text, sentiment)
98
-
99
  return sentiment
100
 
101
  def upsert(text):
@@ -103,8 +105,6 @@ def upsert(text):
103
  doc_ref = db.collection('Text2SpeechSentimentSave').document(date_time)
104
  doc_ref.set({u'firefield': 'Recognize Speech', u'first': 'https://huggingface.co/spaces/awacke1/TTS-STT-Blocks/', u'last': text, u'born': date_time,})
105
  saved = select('TTS-STT', date_time)
106
-
107
-
108
  return saved
109
 
110
  def select(collection, document):
@@ -138,11 +138,11 @@ def tts(text: str, model_name: str):
138
  demo = gr.Blocks()
139
  with demo:
140
  audio_file = gr.inputs.Audio(source="microphone", type="filepath")
141
- text = gr.Textbox()
142
  label = gr.Label()
143
- saved = gr.Textbox()
144
- savedAll = gr.Textbox()
145
- TTSchoice = gr.inputs.Radio( label="Pick a TTS Model", choices=MODEL_NAMES, )
146
  audio = gr.Audio(label="Output", interactive=False)
147
 
148
  b1 = gr.Button("Recognize Speech")
 
24
  DATA_FILENAME = "MindfulStory.csv"
25
  DATA_FILE = os.path.join("data", DATA_FILENAME)
26
  HF_TOKEN = os.environ.get("HF_TOKEN")
27
+
28
  # Download dataset repo using hub download
29
  try:
30
  hf_hub_download(
 
48
  context = file.read()
49
 
50
  # Set up cloned dataset from repo for operations
51
+ repo = Repository( local_dir="data", clone_from=DATASET_REPO_URL, use_auth_token=HF_TOKEN)
 
 
52
 
53
+ # set up ASR
54
  asr = pipeline("automatic-speech-recognition", "facebook/wav2vec2-base-960h")
55
 
56
+ # set up TTS
57
  MODEL_NAMES = [
58
  "en/ljspeech/tacotron2-DDC",
59
  "en/ljspeech/glow-tts",
 
63
  "fr/mai/tacotron2-DDC",
64
  "de/thorsten/tacotron2-DCA",
65
  ]
66
+
67
+ # Use Model Manager to load vocoders
68
  MODELS = {}
69
  manager = ModelManager()
70
  for MODEL_NAME in MODEL_NAMES:
 
81
  )
82
  MODELS[MODEL_NAME] = synthesizer
83
 
84
+ # transcribe
85
  def transcribe(audio):
86
  text = asr(audio)["text"]
87
  return text
88
 
89
+ #text classifier
90
  classifier = pipeline("text-classification")
91
 
92
+
93
  def speech_to_text(speech):
94
  text = asr(speech)["text"]
 
95
  #rMem = AIMemory("STT", text)
 
96
  return text
97
 
98
  def text_to_sentiment(text):
99
  sentiment = classifier(text)[0]["label"]
 
100
  #rMem = AIMemory(text, sentiment)
 
101
  return sentiment
102
 
103
  def upsert(text):
 
105
  doc_ref = db.collection('Text2SpeechSentimentSave').document(date_time)
106
  doc_ref.set({u'firefield': 'Recognize Speech', u'first': 'https://huggingface.co/spaces/awacke1/TTS-STT-Blocks/', u'last': text, u'born': date_time,})
107
  saved = select('TTS-STT', date_time)
 
 
108
  return saved
109
 
110
  def select(collection, document):
 
138
  demo = gr.Blocks()
139
  with demo:
140
  audio_file = gr.inputs.Audio(source="microphone", type="filepath")
141
+ text = gr.Textbox(label="Speech to Text")
142
  label = gr.Label()
143
+ saved = gr.Textbox(label="Saved")
144
+ savedAll = gr.Textbox(label="SavedAll")
145
+ TTSchoice = gr.inputs.Radio( label="Pick a Text to Speech Model", choices=MODEL_NAMES, )
146
  audio = gr.Audio(label="Output", interactive=False)
147
 
148
  b1 = gr.Button("Recognize Speech")