From 5334f3aefcf5a54f06e767b2ebfdb22257964df9 Mon Sep 17 00:00:00 2001
From: Wolfang Torres <wolfangtorres@qq.com>
Date: Wed, 24 Jun 2026 20:05:51 +0800
Subject: [PATCH] test auding pauses toaudio genberation

---
 src/anki_hsk_creator/constants.py  |  4 ++++
 src/anki_hsk_creator/proccessor.py | 10 ++++++----
 2 files changed, 10 insertions(+), 4 deletions(-)

diff --git a/src/anki_hsk_creator/constants.py b/src/anki_hsk_creator/constants.py
index 69078b8..fb5755d 100644
--- a/src/anki_hsk_creator/constants.py
+++ b/src/anki_hsk_creator/constants.py
@@ -22,6 +22,10 @@ PHRASES_TYPE = ".phrases"
 DICT_TYPE = ".dictionary"
 DICTATION_TYPE = ".dictation"
 
+# time dalays
+WORD = 0.1
+COMMA = 0.5
+POINT = 1.0
 
 class LANGUAGES:
     """Available laguages for translation"""
diff --git a/src/anki_hsk_creator/proccessor.py b/src/anki_hsk_creator/proccessor.py
index c0f189d..a1a6386 100644
--- a/src/anki_hsk_creator/proccessor.py
+++ b/src/anki_hsk_creator/proccessor.py
@@ -8,7 +8,7 @@ import argostranslate.translate
 import torchaudio
 
 # Local
-from .constants import LANGUAGES
+from .constants import LANGUAGES, WORD, COMMA
 from .utility import CCCEDICT, TTS, DictionaryResult, ProcessFile, TranslationResult
 
 # Constants
@@ -27,7 +27,8 @@ def dictation_process(
     for n, line in enumerate(text_lines):
         line = line.strip()
         line = "  ".join(line.split())
-        line = line.replace("，", "， 。。。 ")
+        audio_line = f"[pause:{WORD}s]".join(line)
+        audio_line = audio_line.replace("，", f"，[pause:{COMMA}s]")
         audio_path = process_file.resources / f"N{n:03n}.wav"
         if not audio_path.exists():
             audio = TTS.MODEL.generate(
@@ -51,11 +52,12 @@ def translator_process(
     for n, line in enumerate(text_lines):
         line = line.strip()
         line = "  ".join(line.split())
-        line = line.replace("，", "， 。。。 ")
+        audio_line = "[pause:0.1s]".join(line)
+        audio_line = audio_line.replace("，", "，[pause:0.5s]")
         audio_path = process_file.resources / f"N{n:03n}.wav"
         if not audio_path.exists():
             audio = TTS.MODEL.generate(
-                f"{line}。", language_id=LANGUAGES.CN, **TTS.DEFAULTS
+                f"{audio_line}。", language_id=LANGUAGES.CN, **TTS.DEFAULTS
             )
             torchaudio.save(audio_path, audio, TTS.MODEL.sr)
         translated = argostranslate.translate.translate(