Spaces:
Sleeping
Sleeping
add app file
Browse files
app.py
ADDED
|
@@ -0,0 +1,131 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
+
import numpy as np
|
| 3 |
+
|
| 4 |
+
def calculate_wer(reference, hypothesis):
|
| 5 |
+
reference_words = reference.split()
|
| 6 |
+
hypothesis_words = hypothesis.split()
|
| 7 |
+
|
| 8 |
+
m = len(reference_words)
|
| 9 |
+
n = len(hypothesis_words)
|
| 10 |
+
|
| 11 |
+
# Initialize DP table
|
| 12 |
+
dp = np.zeros((m+1, n+1), dtype=np.int32)
|
| 13 |
+
|
| 14 |
+
# Base cases
|
| 15 |
+
for i in range(m+1):
|
| 16 |
+
dp[i][0] = i
|
| 17 |
+
for j in range(n+1):
|
| 18 |
+
dp[0][j] = j
|
| 19 |
+
|
| 20 |
+
# Fill DP table
|
| 21 |
+
for i in range(1, m+1):
|
| 22 |
+
for j in range(1, n+1):
|
| 23 |
+
cost = 0 if reference_words[i-1] == hypothesis_words[j-1] else 1
|
| 24 |
+
dp[i][j] = min(dp[i-1][j] + 1, # Deletion
|
| 25 |
+
dp[i][j-1] + 1, # Insertion
|
| 26 |
+
dp[i-1][j-1] + cost) # Substitution or no cost
|
| 27 |
+
|
| 28 |
+
wer = dp[m][n] / m
|
| 29 |
+
return wer
|
| 30 |
+
|
| 31 |
+
def calculate_cer(reference, hypothesis):
|
| 32 |
+
reference = reference.replace(" ", "")
|
| 33 |
+
hypothesis = hypothesis.replace(" ", "")
|
| 34 |
+
|
| 35 |
+
m = len(reference)
|
| 36 |
+
n = len(hypothesis)
|
| 37 |
+
|
| 38 |
+
# Initialize DP table
|
| 39 |
+
dp = np.zeros((m+1, n+1), dtype=np.int32)
|
| 40 |
+
|
| 41 |
+
# Base cases
|
| 42 |
+
for i in range(m+1):
|
| 43 |
+
dp[i][0] = i
|
| 44 |
+
for j in range(n+1):
|
| 45 |
+
dp[0][j] = j
|
| 46 |
+
|
| 47 |
+
# Fill DP table
|
| 48 |
+
for i in range(1, m+1):
|
| 49 |
+
for j in range(1, n+1):
|
| 50 |
+
cost = 0 if reference[i-1] == hypothesis[j-1] else 1
|
| 51 |
+
dp[i][j] = min(dp[i-1][j] + 1, # Deletion
|
| 52 |
+
dp[i][j-1] + 1, # Insertion
|
| 53 |
+
dp[i-1][j-1] + cost) # Substitution or no cost
|
| 54 |
+
|
| 55 |
+
cer = dp[m][n] / m
|
| 56 |
+
return cer
|
| 57 |
+
|
| 58 |
+
def calculate_mer(reference, hypothesis):
|
| 59 |
+
# Custom mission error rate calculation
|
| 60 |
+
# Replace this with your specific logic
|
| 61 |
+
return 0.0
|
| 62 |
+
|
| 63 |
+
def process_files(reference_file, hypothesis_file):
|
| 64 |
+
try:
|
| 65 |
+
with open(reference_file.name, 'r') as f:
|
| 66 |
+
reference_text = f.read()
|
| 67 |
+
|
| 68 |
+
with open(hypothesis_file.name, 'r') as f:
|
| 69 |
+
hypothesis_text = f.read()
|
| 70 |
+
|
| 71 |
+
wer_value = calculate_wer(reference_text, hypothesis_text)
|
| 72 |
+
cer_value = calculate_cer(reference_text, hypothesis_text)
|
| 73 |
+
mer_value = calculate_mer(reference_text, hypothesis_text)
|
| 74 |
+
|
| 75 |
+
return {
|
| 76 |
+
"WER": wer_value,
|
| 77 |
+
"CER": cer_value,
|
| 78 |
+
"MER": mer_value
|
| 79 |
+
}
|
| 80 |
+
except Exception as e:
|
| 81 |
+
return {"error": str(e)}
|
| 82 |
+
|
| 83 |
+
def main():
|
| 84 |
+
with gr.Blocks() as demo:
|
| 85 |
+
gr.Markdown("# ASR Metrics Calculator")
|
| 86 |
+
|
| 87 |
+
with gr.Row():
|
| 88 |
+
reference_file = gr.File(label="Upload Reference File")
|
| 89 |
+
hypothesis_file = gr.File(label="Upload Hypothesis File")
|
| 90 |
+
|
| 91 |
+
with gr.Row():
|
| 92 |
+
reference_preview = gr.Textbox(label="Reference Preview", lines=3)
|
| 93 |
+
hypothesis_preview = gr.Textbox(label="Hypothesis Preview", lines=3)
|
| 94 |
+
|
| 95 |
+
with gr.Row():
|
| 96 |
+
compute_button = gr.Button("Compute Metrics")
|
| 97 |
+
results_output = gr.JSON(label="Results")
|
| 98 |
+
|
| 99 |
+
# Update previews when files are uploaded
|
| 100 |
+
def update_previews(ref_file, hyp_file):
|
| 101 |
+
if ref_file:
|
| 102 |
+
with open(ref_file.name, 'r') as f:
|
| 103 |
+
ref_text = f.read()
|
| 104 |
+
return ref_text[:200] # Show first 200 characters
|
| 105 |
+
if hyp_file:
|
| 106 |
+
with open(hyp_file.name, 'r') as f:
|
| 107 |
+
hyp_text = f.read()
|
| 108 |
+
return hyp_text[:200] # Show first 200 characters
|
| 109 |
+
return "", ""
|
| 110 |
+
|
| 111 |
+
reference_file.change(
|
| 112 |
+
fn=update_previews,
|
| 113 |
+
inputs=[reference_file, hypothesis_file],
|
| 114 |
+
outputs=[reference_preview, hypothesis_preview]
|
| 115 |
+
)
|
| 116 |
+
hypothesis_file.change(
|
| 117 |
+
fn=update_previews,
|
| 118 |
+
inputs=[reference_file, hypothesis_file],
|
| 119 |
+
outputs=[reference_preview, hypothesis_preview]
|
| 120 |
+
)
|
| 121 |
+
|
| 122 |
+
compute_button.click(
|
| 123 |
+
fn=process_files,
|
| 124 |
+
inputs=[reference_file, hypothesis_file],
|
| 125 |
+
outputs=results_output
|
| 126 |
+
)
|
| 127 |
+
|
| 128 |
+
demo.launch()
|
| 129 |
+
|
| 130 |
+
if __name__ == "__main__":
|
| 131 |
+
main()
|