Spaces:
Running
on
CPU Upgrade
Running
on
CPU Upgrade
Commit
·
1412907
1
Parent(s):
e1ef382
try
Browse files- app.py +21 -4
- audio_index.py +3 -1
app.py
CHANGED
@@ -28,22 +28,39 @@ def audio_search(audio_tuple, prompt: str):
|
|
28 |
array = array.astype(np.float32) / 32768.0
|
29 |
|
30 |
rows = audio_embedding_system.search((sample_rate, array))
|
31 |
-
|
|
|
32 |
orig_rows = search(rows)
|
33 |
-
for row in rows:
|
34 |
path = row["path"]
|
35 |
for orig in orig_rows:
|
36 |
orig_row = orig["row"]
|
37 |
-
print(orig_row)
|
38 |
if orig_row["path"] == path:
|
39 |
row["sentence"] = orig_row["sentence"]
|
40 |
row["audio"] = [
|
41 |
"<audio src=" + orig_row["audio"][0]["src"] + " controls />"
|
42 |
]
|
43 |
-
|
44 |
by="distance", ascending=True
|
45 |
)
|
46 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
47 |
sample_text = gr.Textbox(
|
48 |
label="Prompt",
|
49 |
info="Hit Enter to get a prompt from the common voice dataset",
|
|
|
28 |
array = array.astype(np.float32) / 32768.0
|
29 |
|
30 |
rows = audio_embedding_system.search((sample_rate, array))
|
31 |
+
least_similar = audio_embedding_system.search((sample_rate, array), least_similar=True)
|
32 |
+
rows += least_similar
|
33 |
orig_rows = search(rows)
|
34 |
+
for i, row in enumerate(rows):
|
35 |
path = row["path"]
|
36 |
for orig in orig_rows:
|
37 |
orig_row = orig["row"]
|
|
|
38 |
if orig_row["path"] == path:
|
39 |
row["sentence"] = orig_row["sentence"]
|
40 |
row["audio"] = [
|
41 |
"<audio src=" + orig_row["audio"][0]["src"] + " controls />"
|
42 |
]
|
43 |
+
df = pd.DataFrame(rows)[["path", "audio", "sentence", "distance"]].sort_values(
|
44 |
by="distance", ascending=True
|
45 |
)
|
46 |
|
47 |
+
# Define the styling function
|
48 |
+
def style_path_column(col):
|
49 |
+
n = len(col)
|
50 |
+
# Default empty styles
|
51 |
+
styles = [''] * n
|
52 |
+
for i in range(n):
|
53 |
+
# First 5 rows: green background with opacity
|
54 |
+
if i < 5:
|
55 |
+
styles[i] = 'background-color: rgba(0, 255, 0, 0.3)'
|
56 |
+
# Last 3 rows: red background with opacity
|
57 |
+
elif i >= n - 3:
|
58 |
+
styles[i] = 'background-color: rgba(255, 0, 0, 0.3)'
|
59 |
+
return styles
|
60 |
+
|
61 |
+
# Apply the styling to the 'path' column and return the Styler object
|
62 |
+
return df.style.apply(style_path_column, subset=['path'])
|
63 |
+
|
64 |
sample_text = gr.Textbox(
|
65 |
label="Prompt",
|
66 |
info="Hit Enter to get a prompt from the common voice dataset",
|
audio_index.py
CHANGED
@@ -116,7 +116,7 @@ class AudioEmbeddingSystem:
|
|
116 |
|
117 |
return current_index_size
|
118 |
|
119 |
-
def search(self, row: dict | tuple, top_k=5):
|
120 |
"""
|
121 |
Search for similar audio files.
|
122 |
Either provide query_audio (path to audio file) or query_embedding (numpy array)
|
@@ -127,6 +127,8 @@ class AudioEmbeddingSystem:
|
|
127 |
query_embedding = get_embedding_from_array(*row)
|
128 |
|
129 |
query_embedding = query_embedding.reshape(1, -1).astype(np.float32)
|
|
|
|
|
130 |
|
131 |
distances, indices = self.index.search(query_embedding, top_k)
|
132 |
|
|
|
116 |
|
117 |
return current_index_size
|
118 |
|
119 |
+
def search(self, row: dict | tuple, top_k=5, least_similar=False):
|
120 |
"""
|
121 |
Search for similar audio files.
|
122 |
Either provide query_audio (path to audio file) or query_embedding (numpy array)
|
|
|
127 |
query_embedding = get_embedding_from_array(*row)
|
128 |
|
129 |
query_embedding = query_embedding.reshape(1, -1).astype(np.float32)
|
130 |
+
if least_similar:
|
131 |
+
query_embedding = -1 * query_embedding
|
132 |
|
133 |
distances, indices = self.index.search(query_embedding, top_k)
|
134 |
|