Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -4,6 +4,29 @@ import torch
|
|
4 |
import librosa
|
5 |
from transformers import AutoTokenizer, AutoModelForSequenceClassification, TextClassificationPipeline, AutoModelForTokenClassification, TokenClassificationPipeline, Wav2Vec2ForCTC, Wav2Vec2Processor, Wav2Vec2ProcessorWithLM
|
6 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
7 |
def greet(name):
|
8 |
return "Hello " + name + "!!"
|
9 |
|
|
|
4 |
import librosa
|
5 |
from transformers import AutoTokenizer, AutoModelForSequenceClassification, TextClassificationPipeline, AutoModelForTokenClassification, TokenClassificationPipeline, Wav2Vec2ForCTC, Wav2Vec2Processor, Wav2Vec2ProcessorWithLM
|
6 |
|
7 |
+
# ASR
|
8 |
+
model_name = "jonatasgrosman/wav2vec2-large-xlsr-53-english"
|
9 |
+
processor_asr = Wav2Vec2Processor.from_pretrained(model_name)
|
10 |
+
model_asr = Wav2Vec2ForCTC.from_pretrained(model_name)
|
11 |
+
|
12 |
+
# Classifier Intent
|
13 |
+
model_name = 'qanastek/XLMRoberta-Alexa-Intents-Classification'
|
14 |
+
tokenizer_intent = AutoTokenizer.from_pretrained(model_name)
|
15 |
+
model_intent = AutoModelForSequenceClassification.from_pretrained(model_name)
|
16 |
+
classifier_intent = TextClassificationPipeline(model=model_intent, tokenizer=tokenizer_intent, device=0)
|
17 |
+
|
18 |
+
# Classifier Language
|
19 |
+
model_name = 'qanastek/51-languages-classifier'
|
20 |
+
tokenizer_langs = AutoTokenizer.from_pretrained(model_name)
|
21 |
+
model_langs = AutoModelForSequenceClassification.from_pretrained(model_name)
|
22 |
+
classifier_language = TextClassificationPipeline(model=model_langs, tokenizer=tokenizer_langs, device=0)
|
23 |
+
|
24 |
+
# NER Extractor
|
25 |
+
model_name = 'qanastek/XLMRoberta-Alexa-Intents-NER-NLU'
|
26 |
+
tokenizer_ner = AutoTokenizer.from_pretrained(model_name)
|
27 |
+
model_ner = AutoModelForTokenClassification.from_pretrained(model_name)
|
28 |
+
predict_ner = TokenClassificationPipeline(model=model_ner, tokenizer=tokenizer_ner, device=0)
|
29 |
+
|
30 |
def greet(name):
|
31 |
return "Hello " + name + "!!"
|
32 |
|