Spaces:
Sleeping
Sleeping
Leo Liu
commited on
Update app.py
Browse files
app.py
CHANGED
@@ -1,58 +1,50 @@
|
|
1 |
-
# import part
|
2 |
-
from transformers import pipeline
|
3 |
import streamlit as st
|
4 |
-
|
5 |
-
|
6 |
-
|
7 |
-
|
8 |
-
|
9 |
-
|
10 |
-
|
11 |
-
|
12 |
-
|
13 |
-
|
14 |
-
|
15 |
-
|
16 |
-
|
17 |
-
#
|
18 |
-
|
19 |
-
|
20 |
-
|
21 |
-
|
22 |
-
|
23 |
-
|
24 |
-
|
25 |
-
|
26 |
-
|
27 |
-
|
28 |
-
|
29 |
-
|
30 |
-
|
31 |
-
|
32 |
-
|
33 |
-
|
34 |
-
|
35 |
-
|
36 |
-
|
37 |
-
|
38 |
-
|
39 |
-
|
40 |
-
|
41 |
-
|
42 |
-
|
43 |
-
#
|
44 |
-
st.
|
45 |
-
|
46 |
-
#
|
47 |
-
|
48 |
-
|
49 |
-
|
50 |
-
|
51 |
-
|
52 |
-
|
53 |
-
if st.button("Play Audio"):
|
54 |
-
#st.audio(audio_data['audio'],
|
55 |
-
# format="audio/wav",
|
56 |
-
# start_time=0,
|
57 |
-
# sample_rate = audio_data['sampling_rate'])
|
58 |
-
st.audio("kids_playing_audio.wav")
|
|
|
|
|
|
|
1 |
import streamlit as st
|
2 |
+
from transformers import pipeline
|
3 |
+
from PIL import Image
|
4 |
+
import soundfile as sf
|
5 |
+
import io
|
6 |
+
|
7 |
+
# 1. 加载Pipeline
|
8 |
+
# - 图像→文本:使用 nlpconnect/vit-gpt2-image-captioning
|
9 |
+
# - 文本→语音:使用 facebook/mms-tts 或其它 TTS 模型
|
10 |
+
img_to_text = pipeline("image-to-text", model="nlpconnect/vit-gpt2-image-captioning")
|
11 |
+
text_to_speech = pipeline("text-to-speech", model="facebook/mms-tts")
|
12 |
+
|
13 |
+
st.title("Image-to-Text and Text-to-Speech App (WAV output)")
|
14 |
+
|
15 |
+
# 2. 上传图片
|
16 |
+
uploaded_image = st.file_uploader("Upload an image", type=["jpg", "png", "jpeg"])
|
17 |
+
if uploaded_image:
|
18 |
+
# 显示图片
|
19 |
+
img = Image.open(uploaded_image)
|
20 |
+
st.image(img, caption="Uploaded Image", use_container_width=True)
|
21 |
+
|
22 |
+
# 3. 图像 → 文本
|
23 |
+
text_output = img_to_text(img)[0]["generated_text"]
|
24 |
+
st.write("### Extracted Text")
|
25 |
+
st.write(text_output)
|
26 |
+
|
27 |
+
# 4. 文本 → 语音 (TTS)
|
28 |
+
# text_to_speech(...) 返回一个 dict,包含 "audio" (numpy数组) 和 "sampling_rate"
|
29 |
+
st.write("### Listen to Speech Output")
|
30 |
+
speech_output = text_to_speech(text_output)
|
31 |
+
|
32 |
+
# 5. 将返回的音频数组写到内存中的 WAV 文件
|
33 |
+
audio_array = speech_output["audio"] # numpy array
|
34 |
+
sample_rate = speech_output["sampling_rate"] # 采样率
|
35 |
+
|
36 |
+
wav_io = io.BytesIO()
|
37 |
+
# 利用 soundfile 将音频数组写入内存,并指定格式为 WAV
|
38 |
+
sf.write(wav_io, audio_array, sample_rate, format="WAV")
|
39 |
+
wav_io.seek(0) # 将指针重置到开头,方便后续读取
|
40 |
+
|
41 |
+
# 6. 使用 st.audio 播放内存中的 WAV
|
42 |
+
st.audio(wav_io, format="audio/wav")
|
43 |
+
|
44 |
+
# 7. (可选) 提供下载按钮,下载 WAV 文件
|
45 |
+
st.download_button(
|
46 |
+
label="Download WAV",
|
47 |
+
data=wav_io,
|
48 |
+
file_name="speech.wav",
|
49 |
+
mime="audio/wav"
|
50 |
+
)
|
|
|
|
|
|
|
|
|
|
|
|