08-21-2048, chapter 4.4-linking-complete

1000-hours/public/jupyter-notebooks/edge-tts-valcab-pronounciation.ipynb

@@ -29,19 +29,10 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 2,
+   "execution_count": 22,
    "id": "71d35cd9",
    "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "pygame 2.6.0 (SDL 2.28.4, Python 3.12.2)\n",
-      "Hello from the pygame community. https://www.pygame.org/contribute.html\n"
-     ]
-    }
-   ],
+   "outputs": [],
    "source": [
     "import edge_tts\n",
     "import os\n",
@@ -73,7 +64,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 3,
+   "execution_count": 20,
    "id": "4146f92e",
    "metadata": {},
    "outputs": [
@@ -82,22 +73,9 @@
      "output_type": "stream",
      "text": [
       "['en-US-GuyNeural', 'en-US-AriaNeural']\n",
-      "apple\n",
-      "../audios/apple-us-male.mp3 exists, skipping...\n",
-      "../audios/apple-us-female.mp3 exists, skipping...\n",
-      "\n",
-      "impossible\n",
-      "../audios/impossible-us-male.mp3 exists, skipping...\n",
-      "../audios/impossible-us-female.mp3 exists, skipping...\n",
-      "\n",
-      "possible\n",
-      "../audios/possible-us-male.mp3 created\n",
-      "../audios/possible-us-female.mp3 created\n",
-      "\n",
-      "ichthyosaur\n",
-      "../audios/ichthyosaur-us-male.mp3 created\n",
-      "../audios/ichthyosaur-us-female.mp3 created\n",
-      "\n"
+      "network\n",
+      "../audios/network-us-male.mp3 created\n",
+      "../audios/network-us-female.mp3 created\n"
      ]
     }
    ],
@@ -114,11 +92,7 @@
     "    print(voices)\n",
     "\n",
     "words = \"\"\"\n",
-    "apple,\n",
-    "impossible,\n",
-    "possible,\n",
-    "ichthyosaur,\n",
-    "\n",
+    "network\n",
     "\"\"\"\n",
     "\n",
     "for word in words.strip().split(','):\n",
@@ -132,7 +106,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 15,
+   "execution_count": 21,
    "id": "2d46cde4",
    "metadata": {},
    "outputs": [
@@ -140,7 +114,7 @@
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "a-woman--without-her--man-is-nothing_openai.mp3\n"
+      "she-will-cherish-those-memories-and-ever-hold-them-close-to-her-heart_openai.mp3\n"
      ]
     }
    ],
@@ -160,7 +134,7 @@
     "        ) as response:\n",
     "            response.stream_to_file(path)\n",
     "        \n",
-    "sentence = \"A woman..., without her..., man is nothing.\"\n",
+    "sentence = \"She will cherish those memories and ever hold them close to her heart.\"\n",
     "\n",
     "# remove all punctuation at the end of sentence,\n",
     "# replace all spaces and punctuations in the sentence with dash\n",
@@ -286,6 +260,52 @@
     "* voice = \"en-CA-ClaraNeural\" (Female)\n",
     "* voice = \"en-CA-LiamNeural\" (Male)"
    ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 38,
+   "id": "7350bdce",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "../audios/ten-cups-us-alloy.mp3\n",
+      "../audios/ten-cups-us-female.mp3\n",
+      "../audios/increase-us-alloy.mp3\n",
+      "../audios/increase-us-female.mp3\n",
+      "../audios/brown-gate-us-alloy.mp3\n",
+      "../audios/brown-gate-us-female.mp3\n"
+     ]
+    }
+   ],
+   "source": [
+    "# generate sentences using edge-tts\n",
+    "sentences = \"\"\"\n",
+    "ten cups\n",
+    "increase\n",
+    "brown gate\n",
+    "\"\"\"\n",
+    "# split the sentences into lines\n",
+    "sentences = [s for s in sentences.strip().split(\"\\n\") if s]\n",
+    "for sentence in sentences:\n",
+    "    # get the first three worrds and replace the first with 'sentence', join them with '-'ArithmeticError\n",
+    "    # filename = f'../audios/sentence-{sentence.split(\" \")[1]}-{sentence.split(\" \")[2]}.mp3'\n",
+    "    # for voice in [\"en-US-GuyNeural\", \"en-US-AriaNeural\"]:\n",
+    "    for voice in [\"alloy\", \"nova\"]:\n",
+    "\n",
+    "        filename = f'../audios/{sentence.rstrip(\",.!?\").replace(\"'\", \"\").replace(\" \", \"-\")}-us-{voice}.mp3'.replace(\"alloy\", \"male\").replace(\"nova\", \"female\")\n",
+    "        # filename = f'../audios/{sentence.rstrip(\",.!?\").replace(\"'\", \"\").replace(\" \", \"-\")}-us-{voice}.mp3'.replace(\"en-US-GuyNeural\", \"male\").replace(\"en-US-AriaNeural\", \"female\")\n",
+    "        print(filename)\n",
+    "        # await generate_edge_tts_audio(sentence, filename, voice=voice, verbose=True, overwrite=True, play=True)\n",
+    "        get_openai_tts_audio(sentence, filename, performer=voice)\n",
+    "    \n",
+    "    \n",
+    "\n",
+    "    # get_openai_tts_audio(sentence, f'../audios/{sentence.replace(\" \", \"-\")}-alloy.mp3', performer='alloy')\n",
+    "    # get_openai_tts_audio(sentence, f'../audios/{sentence.replace(\" \", \"-\")}-nova.mp3', performer='nova')\n"
+   ]
   }
  ],
  "metadata": {

1000-hours/sounds-of-american-english/4.4-linking.md

@@ -58,9 +58,13 @@
 
 ## 4.4.2.2. 叠音（geminates）
 
-上一个音节末尾的辅音和下一个音节开头的辅音相同的时候，这一对相同的辅音被称作**叠音**（*geminates*）。在连贯的自然语音中，这两个辅音就只发声一次 —— 但听起来的时候，有可能感受到前一个辅音存在造成细微停顿。它从机制上来看，其实和**省音**（[4.4.2.1](2.4.3-cc#_2-4-3-1-省音-elision)）一样。其实哪怕相邻的两个辅音相似的时候，也会发生同样的现象。
+上一个音节末尾的辅音和下一个音节开头的辅音相同的时候，这一对相同的辅音被称作**叠音**（*geminates*）。在连贯的自然语音中，这两个辅音就只发声一次 —— 但听起来的时候，有可能感受到前一个辅音存在造成细微停顿。它从机制上来看，其实和**省音**一样。其实哪怕相邻的两个辅音相似的时候，也会发生同样的现象。
 
-比如，*that time*, 不是 <span class="pho">ðæt taɪm/</span>，而是 <span class="pho alt">ðæ·t̬aɪm</span><span class="speak-word-inline" data-audio-us-male="/audios/at-that-time-us-male.mp3" data-audio-us-female="/audios/at-that-time-us-female.mp3"></span> —— 第一个 <span class="pho">t</span> 由于后面跟着一个辅音，于是产生了**省音**，听不到了，但，该有的停顿却依然在，在这里我们用 <span class="pho">·</span> 代表那个停顿 ；而后，第二个 <span class="pho">t</span> 与第三个 <span class="pho">t</span> 是**叠音**，这两个 <span class="pho">t</span> 只发声一次。又比如，*bad day*, 不是 <span class="pho alt">bæd deɪ</span>，而是 <span class="pho alt">bæ·deɪ</span><span class="speak-word-inline" data-audio-us-male="/audios/a-bad-day-us-male.mp3" data-audio-us-female="/audios/a-bad-day-us-female.mp3"></span>。 再比如，*with this*, 不是 <span class="pho alt">wɪθ ðɪs</span>，而是 <span class="pho alt">wɪ·ðɪs</span><span class="speak-word-inline" data-audio-us-male="/audios/with-this-us-male.mp3" data-audio-us-female="/audios/with-this-us-female.mp3"></span>。
+比如，*at that time*, 不是 <span class="pho alt">æt ðæt taɪm</span>，而是 <span class="pho alt">æ·ðæ·t̬aɪm</span><span class="speak-word-inline" data-audio-us-male="/audios/at-that-time-us-male.mp3" data-audio-us-female="/audios/at-that-time-us-female.mp3"></span> —— 第一个 <span class="pho">t</span> 由于后面跟着一个辅音，于是产生了**省音**，听不到了，但，该有的停顿却依然在，在这里我们用 <span class="pho">·</span> 代表那个停顿 ；而后，第二个 <span class="pho">t</span> 与第三个 <span class="pho">t</span> 是**叠音**，这两个 <span class="pho">t</span> 只发声一次。
+
+上下居中的圆点 <span class="pho">·</span> 是我们用来标记这种因为停顿（stop）的符号。（注意，不是英文句号 <span class="pho">.</span> —— 那是用来标记音节边界的符号。）
+
+又比如，*a bad day*, 不是 <span class="pho alt">ə bæd deɪ</span>，而是 <span class="pho alt">ə bæ·deɪ</span><span class="speak-word-inline" data-audio-us-male="/audios/a-bad-day-us-male.mp3" data-audio-us-female="/audios/a-bad-day-us-female.mp3"></span>。 再比如，*with this*, 不是 <span class="pho alt">wɪθ ðɪs</span>，而是 <span class="pho alt">wɪ·ðɪs</span><span class="speak-word-inline" data-audio-us-male="/audios/with-this-us-male.mp3" data-audio-us-female="/audios/with-this-us-female.mp3"></span>。
 
 ## 4.4.2.3. 同化（assimilation）
 
@@ -69,18 +73,24 @@
 > * <span class="pho">t</span> + <span class="pho">j</span> = <span class="pho">tʃ</span>
 > * <span class="pho">d</span> + <span class="pho">j</span> = <span class="pho">dʒ</span>
 
-最常见的比如，*Don't you?* <span class="pho alt">doʊn tʃju?</span><span class="speak-word-inline" data-audio-us-male="/audios/Dont-you-us-male.mp3" data-audio-us-female="/audios/Dont-you-us-female.mp3"></span>；以及，*Would you?* <span class="pho alt">wʊ dʒju?</span><span class="speak-word-inline" data-audio-us-male="/audios/Would-you-us-male.mp3" data-audio-us-female="/audios/Would-you-us-female.mp3"></span> 这里的关键在于，<span class="pho">t, d</span> 发声时起始的舌尖位置如果放对了（[2.2.3](2.2.3-td)），它与后面的 <span class="pho alt">ju</span> 连起来的发声就只能如此。
+最常见的比如，*Don't you?* <span class="pho alt">doʊn tʃju?</span><span class="speak-word-inline" data-audio-us-male="/audios/Dont-you-us-male.mp3"></span> *Would you?* <span class="pho alt">wʊ dʒju?</span><span class="speak-word-inline" data-audio-us-female="/audios/Would-you-us-female.mp3"></span> 这里的关键在于，<span class="pho">t, d</span> 发声时起始的舌尖位置如果放对了（[2.2.3](2.2.3-td)），它与后面的 <span class="pho alt">ju</span> 连起来的发声就只能如此。
 
 同化主要发生在 <span class="pho">j</span> 这个音上。除了已经讲过的 <span class="pho">j</span> 前面遇到 <span class="pho">t</span> 和 <span class="pho">d</span> 时会产生同化之外，
 
 > * <span class="pho">s</span> + <span class="pho">j</span> = <span class="pho">ʃ</span>
 > * <span class="pho">z</span> + <span class="pho">j</span> = <span class="pho">ʒ</span>
 
-比如，<span class="pho">s+j</span> 可能会被读成 <span class="pho">ʃ</span>，<span class="pho">z+j</span> 可能会被读成 <span class="pho">ʒ</span>。再比如，*Miss you so much.* <span class="pho alt">mɪ ʃu səʊ mʌʧ.</span><span class="speak-word-inline" data-audio-us-male="/audios/Miss-you-so-much-us-male.mp3" data-audio-us-female="/audios/Miss-you-so-much-us-female.mp3"></span>；*It was your job.* <span class="pho alt">ɪt wɒ ʒʊr ʤɒb.</span><span class="speak-word-inline" data-audio-us-male="/audios/It-was-your-job-us-male.mp3" data-audio-us-female="/audios/It-was-your-job-us-female.mp3"></span>
+比如，<span class="pho">s+j</span> 可能会被读成 <span class="pho">ʃ</span>，<span class="pho">z+j</span> 可能会被读成 <span class="pho">ʒ</span>。再比如，*Miss you so much.* <span class="pho alt">mɪ ʃu səʊ mʌʧ.</span><span class="speak-word-inline" data-audio-us-male="/audios/Miss-you-so-much-us-male.mp3" data-audio-us-female="/audios/Miss-you-so-much-us-female.mp3"></span>*It was your job.* <span class="pho alt">ɪt wɒ ʒʊr ʤɒb.</span><span class="speak-word-inline" data-audio-us-male="/audios/It-was-your-job-us-male.mp3" data-audio-us-female="/audios/It-was-your-job-us-female.mp3"></span>
 
-请注意，这两组是**可能**，并非一定，并非所有英文母语使用者都如此发音。
+另外一个细节是，<span class="pho">k/g</span> 之前的 <span class="pho">n</span>，**可能**会被同化为 <span class="pho">ŋ</span>:
 
-另外一个细节是，<span class="pho">k/g</span> 之前的 <span class="pho">n</span>，可能会被同化为 <span class="pho">ŋ</span>:
+> * *ten cups*, <span class="pho alt">ten kʌps</span> → <span class="pho alt">teŋ kʌps</span><span class="speak-word-inline" data-audio-us-male="/audios/ten-cups-us-male.mp3" data-audio-us-female="/audios/ten-cups-us-female.mp3"></span>
+> * *increase*, <span class="pho alt">ˌɪnˈkris</span> → <span class="pho alt">ˌɪŋˈkris</span><span class="speak-word-inline" data-audio-us-male="/audios/increase-us-male.mp3" data-audio-us-female="/audios/increase-us-female.mp3"></span>
+> * *brown gate*, <span class="pho alt">braʊn geɪt</span> → <span class="pho alt">braʊŋ geɪt</span><span class="speak-word-inline" data-audio-us-male="/audios/brown-gate-us-male.mp3" data-audio-us-female="/audios/brown-gate-us-female.mp3"></span>
+
+> [!Note]
+> 
+> 请注意，**同化**都是**可能**，并非一定，并非所有英文母语使用者都如此发音。并非所有人都会使用这种**同化**的发音。比如，*Don't you?* <span class="pho alt">doʊn tʃju?</span><span class="speak-word-inline" data-audio-us-female="/audios/Dont-you-us-female.mp3"></span> *Would you?* <span class="pho alt">wʊ dʒju?</span><span class="speak-word-inline" data-audio-us-male="/audios/Would-you-us-male.mp3"></span>
 
 ## 4.4.2.4. 击穿（dropping）
 
@@ -90,7 +100,7 @@
 
 很多美国人在 <span class="pho">t</span> 这个音上还有另外一个特殊的习惯，<span class="pho">t</span> 跟在 <span class="pho">n</span> 之后的时候，美国人说话的方式会让我们听不到那个 <span class="pho">t</span>，也就是说，<span class="pho">t</span> 也可能被击穿。比如，他们说 *interview* 的时候你听到的可能是 <span class="pho alt">ˈɪnəˌvju</span><span class="speak-word-inline" data-audio-us-male="/audios/interview-us-male.mp3" data-audio-us-female="/audios/interview-us-female.mp3"></span>，没有 <span class="pho">t</span>。
 
-另外，<span class="pho">ð</span> 也可能被击穿，基本只发生在 *them* 这个常用词上。比如，*let them* <span class="pho alt">let ðəm</span>, 变成了 <span class="pho alt">let-əm</span>，又由于 <span class="pho">ð</span> 被击穿了之后，<span class="pho">t</span> 等于被夹在了两个元音之间，于是会变成浊化的弹舌音 <span class="pho">t̬</span>，于是，说出来的或者听到的是 <span class="pho alt">let̬əm</span>…… 所以，在快速的自然语流中 *let him* 和 *let them* 实际上几乎没有区别，需要通过上下文确定究竟是 *him* 还是 *them*。
+另外，<span class="pho">ð</span> 也可能被击穿，基本只发生在 *them* 这个常用词上。比如，*let them* <span class="pho alt">let ðəm</span>, 变成了 <span class="pho alt">let-əm</span>，又由于 <span class="pho">ð</span> 被击穿了之后，<span class="pho">t</span> 等于被夹在了两个元音之间，于是会变成浊化的弹舌音 <span class="pho">t̬</span>，于是，说出来的或者听到的是 <span class="pho alt">let̬əm</span><span class="speak-word-inline" data-audio-us-male="/audios/let-them-us-male.mp3" data-audio-us-female="/audios/let-them-us-female.mp3"></span>…… 所以，在快速的自然语流中 *let him* 和 *let them* 实际上几乎没有区别，需要通过上下文确定究竟是 *him* 还是 *them*。
 
 ## 4.4.2.5 其它（others）
 
@@ -107,7 +117,7 @@
 > - *She is* - <span class="pho alt">ˈʃiː ˈɪz</span> ⭢ <span class="pho alt">ˈʃiːʲɪz</span><span class="speak-word-inline" data-audio-us-male="/audios/She-is-us-male.mp3" data-audio-us-female="/audios/She-is-us-female.mp3"></span>
 > - *They are* - <span class="pho alt">ˈðeɪ ˈɑːr</span> ⭢ <span class="pho alt">ˈðeɪʲɑːr</span><span class="speak-word-inline" data-audio-us-male="/audios/They-are-us-male.mp3" data-audio-us-female="/audios/They-are-us-female.mp3"></span>
 
-注意：一些以 <span class="pho">i</span> 结尾的动词，使用进行时态的时候，末尾会加上 *ing*，即，<span class="pho alt">ɪŋ</span>。比如，*study* <span class="pho alt">ˈstʌdi</span> 的现在分词形式是 *studying* <span class="pho alt">ˈstʌdi.ɪŋ</span> —— 这里会出现加音 <span class="pho">ʲ</span>，即最后一个音节变成 <span class="pho alt">jɪŋ</span>…… 可是，在自然读音中，它更接近 <span class="pho alt">jəŋ</span>，所以，*studying* 的读音，词典里标注的当然还是 <span class="pho alt">ˈstʌdiɪŋ</span>，但，读出来的更接近 <span class="pho alt">ˈstʌdiʲəŋ</span>。
+注意：一些以 <span class="pho">i</span> 结尾的动词，使用进行时态的时候，末尾会加上 *ing*，即，<span class="pho alt">ɪŋ</span>。比如，*study* <span class="pho alt">ˈstʌdi</span> 的现在分词形式是 *studying* <span class="pho alt">ˈstʌdi.ɪŋ</span><span class="speak-word-inline" data-audio-us-male="/audios/studying-us-male.mp3" data-audio-us-female="/audios/studying-us-female.mp3"></span> —— 这里会出现加音 <span class="pho">ʲ</span>，即最后一个音节变成 <span class="pho alt">jɪŋ</span> —— 虽然，*studying* 在词典里标注的只能是 <span class="pho alt">ˈstʌdiɪŋ</span>。
 
 再比如：