Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -1,4 +1,5 @@
|
|
1 |
import gradio as gr
|
|
|
2 |
from transformers import Wav2Vec2Processor, Wav2Vec2ForCTC
|
3 |
import torch
|
4 |
import librosa
|
@@ -8,6 +9,8 @@ processor = Wav2Vec2Processor.from_pretrained("facebook/wav2vec2-large-960h")
|
|
8 |
model = Wav2Vec2ForCTC.from_pretrained("facebook/wav2vec2-large-960h")
|
9 |
|
10 |
def transcribe_speech(audio_path):
|
|
|
|
|
11 |
speech, _ = librosa.load(audio_path, sr=16000)
|
12 |
input_values = processor(speech, return_tensors="pt", padding="longest").input_values
|
13 |
with torch.no_grad():
|
@@ -17,6 +20,8 @@ def transcribe_speech(audio_path):
|
|
17 |
return transcription[0]
|
18 |
|
19 |
def get_dreamtalk(image_in, speech):
|
|
|
|
|
20 |
try:
|
21 |
client = Client("https://fffiloni-dreamtalk.hf.space/")
|
22 |
result = client.predict(
|
@@ -31,6 +36,8 @@ def get_dreamtalk(image_in, speech):
|
|
31 |
raise gr.Error(f"Error in get_dreamtalk: {str(e)}")
|
32 |
|
33 |
def pipe(text, voice, image_in):
|
|
|
|
|
34 |
try:
|
35 |
speech = transcribe_speech(voice)
|
36 |
video = get_dreamtalk(image_in, speech)
|
|
|
1 |
import gradio as gr
|
2 |
+
from gradio_client import Client
|
3 |
from transformers import Wav2Vec2Processor, Wav2Vec2ForCTC
|
4 |
import torch
|
5 |
import librosa
|
|
|
9 |
model = Wav2Vec2ForCTC.from_pretrained("facebook/wav2vec2-large-960h")
|
10 |
|
11 |
def transcribe_speech(audio_path):
|
12 |
+
if audio_path is None:
|
13 |
+
raise gr.Error("No audio file provided.")
|
14 |
speech, _ = librosa.load(audio_path, sr=16000)
|
15 |
input_values = processor(speech, return_tensors="pt", padding="longest").input_values
|
16 |
with torch.no_grad():
|
|
|
20 |
return transcription[0]
|
21 |
|
22 |
def get_dreamtalk(image_in, speech):
|
23 |
+
if image_in is None or speech is None:
|
24 |
+
raise gr.Error("Image or speech input is missing.")
|
25 |
try:
|
26 |
client = Client("https://fffiloni-dreamtalk.hf.space/")
|
27 |
result = client.predict(
|
|
|
36 |
raise gr.Error(f"Error in get_dreamtalk: {str(e)}")
|
37 |
|
38 |
def pipe(text, voice, image_in):
|
39 |
+
if text is None or voice is None or image_in is None:
|
40 |
+
raise gr.Error("All inputs (text, voice, image) are required.")
|
41 |
try:
|
42 |
speech = transcribe_speech(voice)
|
43 |
video = get_dreamtalk(image_in, speech)
|