From d864da4d65fdfb05e24db79fe67e72093d2adf7c Mon Sep 17 00:00:00 2001
From: Triplecq <triplecq@gmail.com>
Date: Thu, 25 Jan 2024 01:25:28 +0900
Subject: [PATCH] validation scripts

---
 egs/reazonspeech/ASR/decode_greedy.sh        |  7 ++++---
 egs/reazonspeech/ASR/decode_modified_beam.sh | 11 ++++++-----
 egs/reazonspeech/ASR/summary.sh              | 16 ++++++++++++++++
 3 files changed, 26 insertions(+), 8 deletions(-)
 create mode 100755 egs/reazonspeech/ASR/summary.sh

diff --git a/egs/reazonspeech/ASR/decode_greedy.sh b/egs/reazonspeech/ASR/decode_greedy.sh
index 159341d61..4f24d0a4e 100755
--- a/egs/reazonspeech/ASR/decode_greedy.sh
+++ b/egs/reazonspeech/ASR/decode_greedy.sh
@@ -1,5 +1,5 @@
 num_epochs=30
-for ((i=$num_epochs; i>=1; i--));
+for ((i=$num_epochs; i>=15; i--));
 do
     for ((j=1; j<=$i; j++));
     do
@@ -7,8 +7,9 @@ do
             --epoch $i \
             --avg $j \
             --exp-dir zipformer/exp \
-            --max-duration 300 \
+            --max-duration 450 \
             --lang data/lang_char \
-            --decoding-method greedy_search
+            --decoding-method greedy_search \
+	    --blank-penalty 2
     done
 done
diff --git a/egs/reazonspeech/ASR/decode_modified_beam.sh b/egs/reazonspeech/ASR/decode_modified_beam.sh
index 3fd91de32..a3a0c25a0 100755
--- a/egs/reazonspeech/ASR/decode_modified_beam.sh
+++ b/egs/reazonspeech/ASR/decode_modified_beam.sh
@@ -1,14 +1,15 @@
 num_epochs=30
-for ((i=$num_epochs; i>=1; i--));
+for ((i=$num_epochs; i>=20; i--));
 do
-    for ((j=1; j<=$i; j++));
+    for avg in 12 11 10 9 8 7 6 5;
     do
         python3 ./zipformer/decode.py \
             --epoch $i \
-            --avg $j \
+            --avg $avg \
             --exp-dir zipformer/exp \
-            --max-duration 300 \
+            --max-duration 450 \
             --lang data/lang_char \
-            --decoding-method modified_beam_search
+            --decoding-method modified_beam_search \
+	    --blank-penalty 2.5
     done
 done
diff --git a/egs/reazonspeech/ASR/summary.sh b/egs/reazonspeech/ASR/summary.sh
new file mode 100755
index 000000000..3f708aeff
--- /dev/null
+++ b/egs/reazonspeech/ASR/summary.sh
@@ -0,0 +1,16 @@
+#!/bin/bash
+
+summary_file="summary_results.txt"
+> "$summary_file"
+
+for file in wer-summary-valid-*-epoch-*-avg-*.txt; do
+    # Extract epoch and avg values from the filename
+    epoch=$(echo "$file" | grep -oP 'epoch-\K[0-9]+')
+    avg=$(echo "$file" | grep -oP 'avg-\K[0-9]+')
+
+    # Extract the WER value from the second line of the file
+    wer=$(awk 'NR==2 {print $2}' "$file")
+
+    # Append the results to the summary file
+    echo "epoch $epoch avg $avg greedy_search $wer" >> "$summary_file"
+done