08-21-1355, chapter 4.3-group

1000-hours/.vitepress/config.mts

@@ -226,8 +226,12 @@ export default withMermaid(
                     link: "/sounds-of-american-english/4.3-group",
                   },
                   {
-                    text: "4.4. 句子",
-                    link: "/sounds-of-american-english/4.4-sentences",
+                    text: "4.4. 连接",
+                    link: "/sounds-of-american-english/4.4-linking",
+                  },
+                  {
+                    text: "4.5. 句子",
+                    link: "/sounds-of-american-english/4.5-sentences",
                   },
                 ],
               },

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

@@ -29,7 +29,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 49,
+   "execution_count": 2,
    "id": "71d35cd9",
    "metadata": {},
    "outputs": [
@@ -37,19 +37,8 @@
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "['en-US-GuyNeural', 'en-US-AriaNeural']\n",
-      "hello\n",
-      "../audios/hello-us-male.mp3 created\n",
-      "../audios/hello-us-female.mp3 created\n",
-      "\n",
-      "heat\n",
-      "../audios/heat-us-male.mp3 created\n",
-      "../audios/heat-us-female.mp3 created\n",
-      "\n",
-      "high\n",
-      "../audios/high-us-male.mp3 created\n",
-      "../audios/high-us-female.mp3 created\n",
-      "\n"
+      "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"
      ]
     }
    ],
@@ -84,10 +73,34 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 3,
    "id": "4146f92e",
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "name": "stdout",
+     "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"
+     ]
+    }
+   ],
    "source": [
     "\n",
     "voices = [\"en-US-GuyNeural\", \"en-US-AriaNeural\", \"en-GB-RyanNeural\", \"en-GB-LibbyNeural\"]\n",
@@ -101,9 +114,11 @@
     "    print(voices)\n",
     "\n",
     "words = \"\"\"\n",
-    "hello,\n",
-    "heat,\n",
-    "high,\n",
+    "apple,\n",
+    "impossible,\n",
+    "possible,\n",
+    "ichthyosaur,\n",
+    "\n",
     "\"\"\"\n",
     "\n",
     "for word in words.strip().split(','):\n",
@@ -117,10 +132,18 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 63,
+   "execution_count": 15,
    "id": "2d46cde4",
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "a-woman--without-her--man-is-nothing_openai.mp3\n"
+     ]
+    }
+   ],
    "source": [
     "def get_openai_tts_audio(text, path, performer='alloy'):\n",
     "        \n",
@@ -137,12 +160,14 @@
     "        ) as response:\n",
     "            response.stream_to_file(path)\n",
     "        \n",
-    "sentence = \"The explanation you gave was clear but I need a more detailed explanation.\"\n",
+    "sentence = \"A woman..., without her..., man is nothing.\"\n",
     "\n",
     "# remove all punctuation at the end of sentence,\n",
     "# replace all spaces and punctuations in the sentence with dash\n",
-    "audio_filename_openai = sentence.translate(str.maketrans(' ,.?!', '-----')).strip().replace(\"--\", \"-\") + '_openai.mp3'\n",
-    "audio_filename_msedge = sentence.translate(str.maketrans(' ,.?!', '-----')).strip().replace(\"--\", \"-\") + '_msedge.mp3'\n",
+    "audio_filename_openai = sentence.strip().translate(str.maketrans(',.?! ', '-----')).replace(\"--\", \"-\").lower().rstrip('-') + '_openai.mp3'\n",
+    "audio_filename_msedge = sentence.strip().translate(str.maketrans(',.?! ', '-----')).replace(\"--\", \"-\").lower().rstrip('-') + '_msedge.mp3'\n",
+    "\n",
+    "print(audio_filename_openai)\n",
     "# get_openai_tts_audio(sentence, audio_filename_openai, performer='alloy')\n",
     "# await generate_edge_tts_audio(sentence, audio_filename_msedge, voice=\"en-US-GuyNeural\", verbose=True, overwrite=True, play=True)\n",
     "\n",

1000-hours/sounds-of-american-english/4.3-group.md

@@ -1,4 +1,71 @@
 # 4.2. 意群
 
-在自然语流中，很多句子并不是 “一口气” 读完的 —— 句子常常会被分割为若干个**意群**。
+在自然语流中，很多句子并不是 “一口气” 读完的…… 
 
+## 4.2.1. 意群
+
+**句子**，只要略微长一点，就可能会被分割为若干个**意群**。
+
+先听听以下视频里的这段话，同时观察一下语音的 “波谱”（*waveform*）及其 “音高变化”（*pitch contour*）：
+
+> Yet, it is a fact of life that an unlettered peasant is considered ignorant.[^1]
+> <video controls width="720"> <source src="/videos/yet-it-is-a-fact-of-life.mp4" type="video/mp4"></source>Your browser does not support the video tag. </video>
+
+从波谱可以清晰地看到，这一整句话，显然被分成了 3 个**意群**：
+
+> * Yet,
+> * it is a fact of life
+> * that an unlettered peasant is considered ignorant.
+
+> [!Note]
+> 这个视频截取自 [Enjoy App](https://1000h.org/enjoy-app/)。Enjoy 是一个英语学习工具（及其社区）。我们可以用它分析任何语音文件，清楚地看到 “停顿” 或者 “声调” 之类过去我们仅用听力难以完整分别的细节……
+
+事实上，如果你反复多听几遍，你会发现，*peasant* 这个词，<span class="pho alt">ˈpez.ənt</span> 是这样读出来的：<span class="pho alt">ˈpez.ən</span> 之后有个很细微的停顿（stop），而后 <span class="pho">t</span> 与后面的 <span class="pho">ɪz</span> 连了起来…… 本质上，这是因为这个 <span class="pho">t</span> 的舌尖动作完整，然后有个停顿，而后接着说出了后面的 <span class="pho">ɪz</span>，所以它们才听起来被连起来了。
+
+无论是**暂停**（pause）还是**停顿**（stop），虽然它们没有声音，却都在自然语流中都起着重要的作用。
+
+为了将来的说明清楚与方便，我们会将意群之间的语流空隙称为 “**停**”（或者 “**暂停**”，即，*pause*），而把词汇内音节之间的语流空隙称为 “**顿**”（或者 “**停顿**”，即，*stop*）。
+
+> * **停**（**暂停**/*pause*）：发生在意群之间，语流空隙时段较长，可以用来换气，不会产生 “连读” 现象
+> * **顿**（**停顿**/*stop*）：发生在音节之间，语流空隙时段很短（但的确存在），不会用来换气，前后可能产生 “连读现象
+
+## 4.2.2. 停/暂停（pause）
+
+暂停的作用就是 “断句” —— 英文和中文一样的，其实任何语言也都是如此，不同的断句，有可能造成意义或者逻辑的变化。
+
+> * Let's eat, kids. <span class="speak-word-inline" data-audio-us-male="/audios/Let's-eat...-kids.-alloy.mp3"></span>
+> * Let's eat kids. <span class="speak-word-inline" data-audio-us-male="/audios/Let's-eat-kids.-alloy.mp3"></span>
+
+或者
+
+> * A woman without her man is nothing. <span class="speak-word-inline" data-audio-us-male="/audios/A-woman-without-her-man-is-nothing.-alloy.mp3"></span>
+> * A woman: without her, man is nothing. <span class="speak-word-inline" data-audio-us-female="/audios/A-woman,-without-her,-man-is-nothing.-nova.mp3"></span>
+
+而同一意群中细微的**停顿**（*stop*）影响自然语流的节奏，不符合惯常节奏的声音，总是需要 “多一层的处理” 才能够理解。
+
+## 4.2.3. 顿/停顿（stop）
+
+**暂停/停**（*pause*）往往很容易分辨，但，**停顿**（*stop*）就是几乎被所有初学者完全忽视的重要细节。
+
+一个常见的停顿来自于**重读音节**。当我们想要强调某个音节的时候，为了能把它读得相对较重，在此之前往往需要**停顿**一下 —— 做个准备。
+
+比如，你可以试着读一下 *absolutely* <span class="pho alt">ˌæbsəˈluːtli</span><span class="speak-word-inline" data-audio-us-male="/audios/absolutely-us-male.mp3" data-audio-us-female="/audios/absolutely-us-female.mp3"></span>…… 为了把 <span class="pho alt">luː</span> 这个音节读成**重音**（顺带说，这是个长元音），你可能就会不由自主地在发出 <span class="pho">sə</span> 之后略微**停顿**一下…… 拿之前的例子 *individual* <span class="pho alt">ˌɪndəˈvɪdʒuəl</span><span class="speak-word-inline" data-audio-us-male="/audios/individual-us-male.mp3" data-audio-us-female="/audios/individual-us-female.mp3"></span> 再试一下，若是在 <span class="pho alt">vɪ</span> 之前略微**停顿**一下，就能相对更为轻松地将它读成**重音**。
+
+另外一个更为常见的**停顿**相对比较微妙，它就在那里，可不仅听不到，就算说话的人就在眼前，也看不到他们停顿时的发声器官的具体动作。
+
+比如，在 *hotdog* <span class="pho alt">ˈhɑːt.dɑːɡ</span><span class="speak-word-inline" data-audio-us-male="/audios/hotdog-us-male.mp3" data-audio-us-female="/audios/hotdog-us-female.mp3"></span>这个词里，<span class="pho">t</span> 的声音是听不到的，但，它的舌尖动作却是**完整**的，所以，能够听到的不是 <span class="pho alt">ˈhɑtˌdɔg</span>，也不是 <span class="pho alt">ˈhɑˌdɔg</span>，而是 <span class="pho alt">ˈhɑt̚ ˌdɔg</span><span class="speak-word-inline" data-audio-us-male="/audios/hotdog-us-male.mp3" data-audio-us-female="/audios/hotdog-us-female.mp3"></span>。你可以再试试 *network* <span class="pho alt">ˈnetˌwərk</span><span class="speak-word-inline" data-audio-us-male="/audios/network-us-male.mp3" data-audio-us-female="/audios/network-us-female.mp3"></span> 这个词 —— 这里面的 <span class="pho">t</span> 也是同样的机制。（注意，也有人用喉塞音 <span class="pho">ʔ</span> 读这这两个词中的 <span class="pho">t</span>。）
+
+这类**停顿**（*stop*），也叫**塞音**，最常见的有以下几种：
+
+> - 唇塞音（*Bilabial stop*）：<span class="pho">p</span>、<span class="pho">b</span>、<span class="pho">m</span>、<span class="pho">f</span>、<span class="pho">v</span> —— *I just cannot hel**p** myself*. <span class="speak-word-inline" data-audio-us-male="/audios/sentence-just-cannot-alloy.mp3" data-audio-us-female="/audios/sentence-just-cannot-nova.mp3"></span> 第四个单词 <span class="pho alt">help</span> 末尾的 <span class="pho">p</span> 就是一个唇塞音。
+> - 齿塞音（*Dental stop*）：<span class="pho">θ</span>、<span class="pho">ð</span> —— *I wish the pa**th** to success was smoother.*<span class="speak-word-inline" data-audio-us-male="/audios/sentence-wish-the-alloy.mp3" data-audio-us-female="/audios/sentence-wish-the-nova.mp3"></span> 第四个单词 <span class="pho alt">pæθ</span> 末尾的 <span class="pho">θ</span> 就是一个齿塞音。
+> - 龈塞音（*Alveolar stop*）: <span class="pho">t</span>、<span class="pho">d</span> —— *He trie**d** to discuss his idea with his boss.*<span class="speak-word-inline" data-audio-us-male="/audios/sentence-tried-to-alloy.mp3" data-audio-us-female="/audios/sentence-tried-to-nova.mp3"></span> 第二个单词 <span class="pho alt">traɪd</span> 末尾的 <span class="pho">d</span> 就是一个龈塞音。
+> - 喉塞音（*Glottal stop*）：<span class="pho">k</span>, <span class="pho">g</span>, <span class="pho">t</span> —— IPA 里用 <span class="pho">ʔ</span>标注。 *I don't thin**k** that is true.*<span class="speak-word-inline" data-audio-us-male="/audios/sentence-don't-think-alloy.mp3" data-audio-us-female="/audios/sentence-don't-think-nova.mp3"></span> 第三个单词 *think* 末尾的 <span class="pho">k</span> 就是一个喉塞音，<span class="pho alt">θɪŋʔ ðæt</span>。 *uh-oh* <span class="pho alt">ɑʔoʊ</span><span class="speak-word-inline" data-audio-us-female="/audios/Uh-Oh.mp3"></span> 是另外一个例子…… <span class="pho">t</span> 也有可能用这个喉塞音，比如，*utmost* <span class="pho alt">ˈʌʔˌmoʊst</span><span class="speak-word-inline" data-audio-us-male="/audios/utmost-us-male.mp3" data-audio-us-female="/audios/utmost-us-female.mp3"></span>，或者 <span class="pho alt">ˈneʔwɝːk</span><span class="speak-word-inline" data-audio-us-male="/audios/network-us-male.mp3" data-audio-us-female="/audios/network-us-female.mp3"></span>。
+
+不管是哪一种，都是因为那个辅音**虽然并未发出声音，但整个发声器官却动作完整**而形成的。
+
+这是一个非常重要的建议：
+
+> 在研究 “**连读**” 之前，可能更应该先仔细研究 “**停**与**顿**”。
+
+[^1]: 这段语音截取自 [Thomas Sowell](https://en.wikipedia.org/wiki/Thomas_Sowell) 的 [Knowledge and Decisions](https://www.amazon.com/Knowledge-Decisions-Thomas-Sowell-ebook/dp/B09NKRSC3W/) 有声版（Audible）