Spaces:
Build error
Build error
update
Browse files
app.py
CHANGED
@@ -22,10 +22,10 @@ coquiTTS = CoquiTTS()
|
|
22 |
|
23 |
|
24 |
# Driver function
|
25 |
-
def
|
26 |
text1, lang = whisper_stt(audio)
|
27 |
#text1 = model.transcribe(audio)["text"]
|
28 |
-
text2 = lang_model_response(text1)
|
29 |
speech = tts(text2, lang) #'en')
|
30 |
return text1, text2, speech
|
31 |
|
@@ -55,16 +55,17 @@ def whisper_stt(audio):
|
|
55 |
|
56 |
|
57 |
# LLM - Bloom Response
|
58 |
-
def lang_model_response(prompt):
|
59 |
print(f"*****Inside lang_model_response - Prompt is :{prompt}")
|
60 |
p = """Question: How are you doing today?
|
61 |
Answer: I am doing good, thanks.
|
62 |
Question: """
|
63 |
-
prompt = p + prompt + "\n" + "Answer: "
|
64 |
-
#prompt = prompt + "\n"
|
65 |
-
|
66 |
if len(prompt) == 0:
|
67 |
-
prompt = """Can you help me please?
|
|
|
|
|
|
|
|
|
68 |
|
69 |
json_ = {"inputs": prompt,
|
70 |
"parameters":
|
@@ -80,27 +81,28 @@ def lang_model_response(prompt):
|
|
80 |
"wait_for_model": True,
|
81 |
},}
|
82 |
response = requests.post(API_URL, headers=headers, json=json_)
|
83 |
-
print(f"Response is : {response}")
|
84 |
output = response.json()
|
85 |
-
#print(f"output is : {output}")
|
86 |
output_tmp = output[0]['generated_text']
|
87 |
-
print(f"
|
88 |
solution = output_tmp.split("Answer: ")[2].split("\n")[0]
|
89 |
-
|
90 |
-
print(f"Final response after splits is: {solution}")
|
91 |
return solution
|
92 |
|
93 |
# Coqui - Text-to-Speech
|
94 |
def tts(text, language):
|
95 |
-
|
96 |
-
|
97 |
-
|
98 |
-
|
|
|
|
|
|
|
99 |
|
100 |
|
101 |
gr.Interface(
|
102 |
title = 'Testing Whisper',
|
103 |
-
fn=
|
104 |
inputs=[
|
105 |
gr.Audio(source="microphone", type="filepath"), #streaming = True,
|
106 |
# "state"
|
|
|
22 |
|
23 |
|
24 |
# Driver function
|
25 |
+
def driver_fun(audio) :
|
26 |
text1, lang = whisper_stt(audio)
|
27 |
#text1 = model.transcribe(audio)["text"]
|
28 |
+
text2 = lang_model_response(text1, lang)
|
29 |
speech = tts(text2, lang) #'en')
|
30 |
return text1, text2, speech
|
31 |
|
|
|
55 |
|
56 |
|
57 |
# LLM - Bloom Response
|
58 |
+
def lang_model_response(prompt, language):
|
59 |
print(f"*****Inside lang_model_response - Prompt is :{prompt}")
|
60 |
p = """Question: How are you doing today?
|
61 |
Answer: I am doing good, thanks.
|
62 |
Question: """
|
|
|
|
|
|
|
63 |
if len(prompt) == 0:
|
64 |
+
prompt = """Question: Can you help me please?
|
65 |
+
Answer: Sure, I am here for you.
|
66 |
+
Question: """
|
67 |
+
|
68 |
+
prompt = p + prompt + "\n" + "Answer: "
|
69 |
|
70 |
json_ = {"inputs": prompt,
|
71 |
"parameters":
|
|
|
81 |
"wait_for_model": True,
|
82 |
},}
|
83 |
response = requests.post(API_URL, headers=headers, json=json_)
|
84 |
+
#print(f"Response is : {response}")
|
85 |
output = response.json()
|
|
|
86 |
output_tmp = output[0]['generated_text']
|
87 |
+
print(f"Bloom API Response is : {output_tmp}")
|
88 |
solution = output_tmp.split("Answer: ")[2].split("\n")[0]
|
89 |
+
print(f"Final Bloom Response after splits is: {solution}")
|
|
|
90 |
return solution
|
91 |
|
92 |
# Coqui - Text-to-Speech
|
93 |
def tts(text, language):
|
94 |
+
print(f"Inside tts - language is : {language}")
|
95 |
+
coqui_langs = ['en' ,'es' ,'fr' ,'de' ,'pl' ,'uk' ,'ro' ,'hu' ,'bg' ,'nl' ,'fi' ,'sl' ,'lv' ,'ga']
|
96 |
+
if language not in coqui_langs:
|
97 |
+
language = 'en'
|
98 |
+
with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as fp:
|
99 |
+
coquiTTS.get_tts(text, fp, speaker = {"language" : language})
|
100 |
+
return fp.name
|
101 |
|
102 |
|
103 |
gr.Interface(
|
104 |
title = 'Testing Whisper',
|
105 |
+
fn=driver_fun,
|
106 |
inputs=[
|
107 |
gr.Audio(source="microphone", type="filepath"), #streaming = True,
|
108 |
# "state"
|