updATE MODELS for audio generation

anki-hsk-creator/__main__.py

@@ -44,34 +44,34 @@ font-size: 75px;
 
 ## Classess
 
-SIMPLE_MODEL = Model(
+PHRASE_MODEL = Model(
     2076166425,
-    "Simple Model",
+    "Phrase Model",
     fields=[
-        {"name": "Question"},
-        {"name": "Answer"},
+        {"name": "Translated"},
+        {"name": "Phrase"},
+        {"name": "Audio"},
     ],
     templates=[
         {
             "name": "Card 1",
-            "qfmt": "{{Question}}",
-            "afmt": '{{FrontSide}}<hr id="answer">{{Answer}}',
+            "qfmt": "{{Translated}}<br>{{Audio}}",
+            "afmt": '{{FrontSide}}<hr id="answer">{{Phrase}}',
+        },
+        {
+            "name": "Card 2",
+            "qfmt": "{{Phrase}}<br>{{Audio}}",
+            "afmt": '{{FrontSide}}<hr id="answer">{{Translated}}',
+        },
+        {
+            "name": "Card 3",
+            "qfmt": "{{Audio}}",
+            "afmt": '{{FrontSide}}<hr id="answer">{{Phrase}}',
         },
     ],
     css=CSS,
 )
 
-HSK_FRONT_TEMPLATE = """
-<tts service="android" voice="zh-CN">
-<strong>{{Pinyin}}</strong>
-</tts>
-<br>
-<tts service="android" voice="en-US">
-{{English}}
-</tts>
-<br>
-{{Audio}}
-"""
 
 HSK_MODEL = Model(
     1708536519,
@@ -81,12 +81,12 @@ HSK_MODEL = Model(
         {"name": "Pinyin"},
         {"name": "Simplified"},
         {"name": "Traditional"},
-        {'name': 'Audio'},
+        {"name": "Audio"},
     ],
     templates=[
         {
             "name": "Card 1",
-            "qfmt": HSK_FRONT_TEMPLATE,
+            "qfmt": "<strong>{{Pinyin}}</strong><br>{{English}}<br>{{Audio}}",
             "afmt": "{{FrontSide}}<hr id='answer''><div class='simple'>"
             "{{Simplified}}</div><br><div class='trad'>{{Traditional}}</div>",
         },
@@ -96,6 +96,12 @@ HSK_MODEL = Model(
             "{{Traditional}}</div>",
             "afmt": '{{FrontSide}}<hr id="answer"><strong>{{Pinyin}}</strong><br>{{English}}<br>{{Audio}}',
         },
+        {
+            "name": "Card 1",
+            "qfmt": "{{Audio}}",
+            "afmt": "{{FrontSide}}<hr id='answer''><strong>{{Pinyin}}</strong><br><div class='simple'>"
+            "{{Simplified}}</div><br><div class='trad'>{{Traditional}}</div>",
+        },
     ],
     css=CSS,
 )
@@ -129,6 +135,7 @@ def create_translator():
     )
     argostranslate.package.install_from_path(package_to_install.download())
 
+
 def create_tts():
     # Automatically detect the best available device
     if torch.cuda.is_available():
@@ -137,9 +144,10 @@ def create_tts():
         device = "mps"
     else:
         device = "cpu"
-    tts = ChatterboxMultilingualTTS.from_pretrained(device=device)
+    tts = ChatterboxMultilingualTTS.from_pretrained(device=device, t3_model="v3")
     return tts
 
+
 ## Main
 
 
@@ -193,8 +201,14 @@ def dictionary_process(dictionary, tts, in_file, resources):
                                 for m in w.meanings:
                                     print(f"\t{m}")
                             s = None
-                            while not s or not s.isnumeric() or not (1 <= int(s) <= len(v)):
-                                s = input(f"Please select the correct word [1-{len(v)}]: ")
+                            while (
+                                not s
+                                or not s.isnumeric()
+                                or not (1 <= int(s) <= len(v))
+                            ):
+                                s = input(
+                                    f"Please select the correct word [1-{len(v)}]: "
+                                )
                             v = v[int(s) - 1]
                         else:
                             v = ml[0]
@@ -202,7 +216,7 @@ def dictionary_process(dictionary, tts, in_file, resources):
                         v = v[0]
                     audio_path = resources / f"{word}.wav"
                     if not audio_path.exists():
-                        audio = tts.generate(word, language_id="zh")
+                        audio = tts.generate(f"{word}。", language_id="zh")
                         torchaudio.save(audio_path, audio, tts.sr)
                     input_file.write(f"{word}\t{v.pinyin}\n")
                     results.append((v, audio_path))
@@ -217,16 +231,18 @@ def dictionary_process(dictionary, tts, in_file, resources):
     return results
 
 
-def translator_process(in_file):
-    """Process text trasnlate files"""
-    text_list = in_file.open(encoding="utf8").read().split()[1:]
+def translator_process(tts, resources, in_file):
+    """Process for phases trasnlation"""
+    text_list = in_file.open(encoding="utf8").read().strip().split()
     results = []
-    for text in text_list:
-        text = text.strip()
-        for par in text.split("。"):
-            if par:
-                translatedText = argostranslate.translate.translate(par, CN, EN)
-                results.append([translatedText, par])
+    for n, phrase in enumerate(text_list):
+        phrase = phrase.strip()
+        audio_path = resources / f"N{n}.wav"
+        if not audio_path.exists():
+            audio = tts.generate(f"{phrase}。", language_id="zh")
+            torchaudio.save(audio_path, audio, tts.sr)
+        translated = argostranslate.translate.translate(phrase, CN, EN)
+        results.append([translated, phrase, audio_path])
     return results
 
 
@@ -260,7 +276,7 @@ def output_anki_dictionary(out_file, results):
                 PinyinToneConverter().convert_text(entry.pinyin),
                 entry.simplified,
                 entry.traditional,
-                f"[sound:{audio.name}]"
+                f"[sound:{audio.name}]",
             ],
         )
         audios.append(audio)
@@ -269,32 +285,41 @@ def output_anki_dictionary(out_file, results):
     package.write_to_file(final_file)
 
 
-def output_anki_text(out_file, results):
+def output_anki_phrase(out_file, results):
     final_file = out_file.parent / f"{out_file.stem}.apkg"
     deck_name = "::".join(out_file.relative_to(OUTPUT).parts[:-1] + (out_file.stem,))
     deck = Deck(random.randrange(1 << 30, 1 << 31), deck_name)
-    for entry in results:
+    package = Package(deck)
+    audios = []
+    for translated, phrase, audio in results:
         note = Note(
-            model=SIMPLE_MODEL,
-            fields=entry,
+            model=PHRASE_MODEL,
+            fields=[
+                translated,
+                phrase,
+                f"[sound:{audio.name}]",
+            ],
         )
         deck.add_note(note)
-    Package(deck).write_to_file(final_file)
+        audios.append(audio)
+    package.media_files = audios
+    package.write_to_file(final_file)
 
 
 def main():
-    in_file, out_file, resources, file_type = process_files()
-    if PHRASES_TYPE in in_file.suffixes:
-        create_translator()
-        results = translator_process(in_file)
-        output_anki_text(out_file, results)
-    elif DICT_TYPE in in_file.suffixes:
-        tts = create_tts()
-        dictionary = create_cedict()
-        results = dictionary_process(dictionary, tts, in_file, resources)
-        output_anki_dictionary(out_file, results)
-    else:
-        raise TypeError("Error, filetype not especified!")
+    tts = create_tts()
+    dictionary = create_cedict()
+    create_translator()
+    while True:
+        in_file, out_file, resources, file_type = process_files()
+        if PHRASES_TYPE in in_file.suffixes:
+            results = translator_process(tts, resources, in_file)
+            output_anki_phrase(out_file, results)
+        elif DICT_TYPE in in_file.suffixes:
+            results = dictionary_process(dictionary, tts, in_file, resources)
+            output_anki_dictionary(out_file, results)
+        else:
+            raise TypeError("Error, filetype not especified!")
 
 
 if __name__ == "__main__":

data/input/HSK1/HSK1-2.dictionary.txt

@@ -1,4 +1,4 @@
-谢谢
-不
-不客气
-再见
+谢谢	xie4 xie5
+不	bu4
+不客气	bu4 ke4 qi5
+再见	zai4 jian4

data/input/HSK1/HSK1-2.phrases.txt

@@ -1,2 +1,7 @@
 不谢
 谢谢你
+打开书
+清大声读
+再读一遍
+一起读
+有问题吗？

data/input/HSK1/HSK1-3.dictionary.txt

@@ -1,12 +1,12 @@
-叫
-什么
-名字
-我
-是
-老师
-吗
-学生
-人
-李明
-中国
-美国
+叫	jiao4
+什么	shen2 me5
+名字	ming2 zi5
+我	wo3
+是	shi4
+老师	lao3 shi1
+吗	ma5
+学生	xue2 sheng5
+人	ren2
+李明	ERROR
+中国	Zhong1 guo2
+美国	Mei3 guo2

data/input/HSK1/HSK1-3.phrases.txt

@@ -0,0 +1,6 @@
+你叫什么名字？
+我叫李月。
+你是老师吗？
+我不是老师， 我是学生。
+你中国人么？
+我不是中国人，我是美国人。

data/input/HSK1/HSK1-4.dictionary.txt

@@ -1,12 +1,12 @@
-她
-他
-谁
-的
-汉语
-哪
-国
-呢
-他
-同学
-朋友
-委内瑞拉
+她	ta1
+他	ta1
+谁	shei2
+的	de5
+汉语	Han4 yu3
+哪	na3
+国	guo2
+呢	ne5
+他	ta1
+同学	tong2 xue2
+朋友	peng2 you5
+委内瑞拉	Wei3 nei4 rui4 la1

data/input/HSK1/HSK1-4.phrases.txt

@@ -0,0 +1,9 @@
+他是谁？
+他是我的汉语老师，他叫李月。
+你是哪国人？
+我是美国人，你呢？
+我是中国人。
+他是谁？
+他是我同学。
+他呢？他是你同学吗？
+他不是我同学，她是我朋友。

data/input/汉语口语新时代-初级下/HSK1-4.phrases.txt

@@ -0,0 +1,9 @@
+他是谁？
+他是我的汉语老师，他叫李月。
+你是哪国人？
+我是美国人，你呢？
+我是中国人。
+他是谁？
+他是我同学。
+他呢？他是你同学吗？
+他不是我同学，她是我朋友。

data/input/汉语口语新时代-初级下/第11课-1.dictionary.txt

@@ -0,0 +1,14 @@
+温柔	wen1 rou2
+星星	xing1 xing5
+摘	zhai1
+颗	ke1
+亲手	qin1 shou3
+浪漫	lang4 man4
+飞行员	fei1 xing2 yuan2
+标准	biao1 zhun3
+考虑	kao3 lu:4
+十全十美	shi2 quan2 shi2 mei3
+理想	li3 xiang3
+支持	zhi1 chi2
+性格	xing4 ge2
+恋爱	lian4 ai4

data/input/汉语口语新时代-初级下/第11课-1.phrases.txt

@@ -0,0 +1,8 @@
+她每年都会送孩子新年礼物
+有一年,孩子要星星,她没办法去亲手,就摘颗送给他。
+有些人在选择男/女朋友时总是有很高的标准,会考虑很多方面。
+但是世界上哪有理想的人啊?
+我觉得十全十美的男/女朋友就是在一起时每天都很开心。
+我理想中的女朋友有两个标准:
+第一,性格要非常温柔;
+第二,要支挡我的工作,即使。