Add medium huayan voice

_script/make_sample_html.py

@@ -34,6 +34,7 @@
     "pt": ("Português", "Portuguese"),
     "pt-br": ("Português (Brasil)", "Portuguese (Brazilian)"),
     "ru": ("Русский", "Russian"),
+    "sv-se": ("Svenska", "Swedish"),
     "sw": "Kiswahili",
     "te": ("తెలుగు", "Telugu"),
     "tn": "Setswana",
@@ -43,6 +44,10 @@
     "zh-cn": ("简体中文", "Chinese"),
 }
 
+LANG_ALIASES = {
+    "zh_CN": "zh-cn",
+}
+
 
 def main():
     with tempfile.NamedTemporaryFile(mode="w+") as output_file, open(
@@ -92,6 +97,7 @@ def write_output(f):
     print('<select id="languages" onchange="jumpLanguage()">', file=f)
     print('<option value="">Jump to language</option>', file=f)
     for language in languages:
+        language = LANG_ALIASES.get(language, language)
         language_name = LANG_NAMES[language]
         if isinstance(language_name, tuple):
             language_name = f"{language_name[0]} ({language_name[1]})"
@@ -123,6 +129,7 @@ def write_output(f):
         print(model_dir)
         quality = model_dir.name
         language = model_dir.parent.parent.name
+        language = LANG_ALIASES.get(language, language)
         assert language in LANG_NAMES, f"No name for {language}"
 
         language_name = LANG_NAMES[language]

_script/test.sh

@@ -0,0 +1,34 @@
+#!/usr/bin/env bash
+set -eo pipefail
+
+this_dir="$( cd "$( dirname "$0" )" && pwd )"
+repo_dir="$(realpath "${this_dir}/../../../../")"
+
+venv="${repo_dir}/src/python/.venv"
+if [ -d "${venv}" ]; then
+    source "${venv}/bin/activate"
+fi
+
+checkpoint="$(find ./ -name '*.ckpt' -type f -printf '%T+ %p\n' | sort -r | head -n1 | cut -d' ' -f2-)"
+sample_rate="$(jq -r '.audio.sample_rate' < "${this_dir}/config.json")"
+language="$(jq -r '.espeak.voice' < "${this_dir}/config.json")"
+num_speakers="$(jq -r '.num_speakers' < "${this_dir}/config.json")"
+
+export PYTHONPATH="${repo_dir}/src/python:${PYTHONPATH}"
+
+if [[ "${num_speakers}" == "1" ]]; then
+    cat "${this_dir}/test_${language}.jsonl" | \
+    python3 -m piper_train.infer \
+        --sample-rate "${sample_rate}" \
+        --checkpoint "${checkpoint}" \
+        --output-dir "${this_dir}/output";
+else
+    for speaker_id in $(seq 0 "$((${num_speakers} - 1))"); do
+        cat "${this_dir}/test_${language}.jsonl" | \
+        jq --compact-output ". += { \"speaker_id\": ${speaker_id} }" | \
+        python3 -m piper_train.infer \
+            --sample-rate "${sample_rate}" \
+            --checkpoint "${checkpoint}" \
+            --output-dir "${this_dir}/output/${speaker_id}";
+    done
+fi

index.html

@@ -88,6 +88,7 @@
 <option value="uk">украї́нська мо́ва (Ukrainian)</option>
 <option value="vi">Tiếng Việt (Vietnamese)</option>
 <option value="zh-cn">简体中文 (Chinese)</option>
+<option value="zh-cn">简体中文 (Chinese)</option>
 </select>
 <hr />
 <h2 id="ca">Català (Catalan)</h2>
@@ -524,6 +525,30 @@ <h2>Dataset</h2>
 <li>License: https://www.cstr.ed.ac.uk/projects/blizzard/2013/lessac_blizzard2013/license.html</li>
 </ul>
 <h2>Training</h2>
+<p>Trained from scratch.</p>
+<div><button value="cancel" formmethod="dialog">Close</button></div>
+</form></dialog>
+<td> <a title="Download en-us-lessac-low" href="https://github.com/rhasspy/piper/releases/download/v0.0.2/voice-en-us-lessac-low.tar.gz">en-us-lessac-low</a> </td>
+<td> <a href="javascript:q('#dialog-en-us-lessac-low').showModal()">View</a> </td>
+<td> low </td>
+<td> <audio id="audio-en-us-lessac-low" preload="none" controls src="samples/en-us-lessac-low/speaker_0/sample.mp3"></audio> <details><summary>sample text</summary><p> A rainbow is a meteorological phenomenon that is caused by reflection, refraction and dispersion of light in water droplets resulting in a spectrum of light appearing in the sky. It takes the form of a multi-colored circular arc. Rainbows caused by sunlight always appear in the section of sky directly opposite the Sun. </p></details> </td>
+<td></td>
+</tr>
+<tr id="en-us-lessac-low" class="">
+<dialog id="dialog-en-us-lessac-low"><form>
+<h1>Model card for Lessac (low)</h1>
+<ul>
+<li>Language: en-us (U.S. English)</li>
+<li>Speakers: 1</li>
+<li>Quality: low</li>
+<li>Samplerate: 16,000Hz</li>
+</ul>
+<h2>Dataset</h2>
+<ul>
+<li>URL: https://www.cstr.ed.ac.uk/projects/blizzard/2013/lessac_blizzard2013/</li>
+<li>License: https://www.cstr.ed.ac.uk/projects/blizzard/2013/lessac_blizzard2013/license.html</li>
+</ul>
+<h2>Training</h2>
 <p>Finetuned from U.S. English Ryan voice (low quality).</p>
 <div><button value="cancel" formmethod="dialog">Close</button></div>
 </form></dialog>
@@ -533,7 +558,7 @@ <h2>Training</h2>
 <td> <audio id="audio-en-us-lessac-low" preload="none" controls src="samples/en-us-lessac-low/speaker_0/sample.mp3"></audio> <details><summary>sample text</summary><p> A rainbow is a meteorological phenomenon that is caused by reflection, refraction and dispersion of light in water droplets resulting in a spectrum of light appearing in the sky. It takes the form of a multi-colored circular arc. Rainbows caused by sunlight always appear in the section of sky directly opposite the Sun. </p></details> </td>
 <td></td>
 </tr>
-<tr id="en-us-lessac-medium" class="">
+<tr id="en-us-lessac-medium" class="alt">
 <dialog id="dialog-en-us-lessac-medium"><form>
 <h1>Model card for Lessac (medium)</h1>
 <ul>
@@ -557,7 +582,7 @@ <h2>Training</h2>
 <td> <audio id="audio-en-us-lessac-medium" preload="none" controls src="samples/en-us-lessac-medium/speaker_0/sample.mp3"></audio> <details><summary>sample text</summary><p> A rainbow is a meteorological phenomenon that is caused by reflection, refraction and dispersion of light in water droplets resulting in a spectrum of light appearing in the sky. It takes the form of a multi-colored circular arc. Rainbows caused by sunlight always appear in the section of sky directly opposite the Sun. </p></details> </td>
 <td></td>
 </tr>
-<tr id="en-us-libritts-high" class="alt">
+<tr id="en-us-libritts-high" class="">
 <dialog id="dialog-en-us-libritts-high"><form>
 <h1>Model card for LibriTTS (high)</h1>
 <ul>
@@ -1486,7 +1511,7 @@ <h2>Training</h2>
 <option value="speaker_903"> 903 </option>
 </select> (904)</td>
 </tr>
-<tr id="en-us-ryan-high" class="">
+<tr id="en-us-ryan-high" class="alt">
 <dialog id="dialog-en-us-ryan-high"><form>
 <h1>Model card for Ryan (high)</h1>
 <ul>
@@ -1510,7 +1535,7 @@ <h2>Training</h2>
 <td> <audio id="audio-en-us-ryan-high" preload="none" controls src="samples/en-us-ryan-high/speaker_0/sample.mp3"></audio> <details><summary>sample text</summary><p> A rainbow is a meteorological phenomenon that is caused by reflection, refraction and dispersion of light in water droplets resulting in a spectrum of light appearing in the sky. It takes the form of a multi-colored circular arc. Rainbows caused by sunlight always appear in the section of sky directly opposite the Sun. </p></details> </td>
 <td></td>
 </tr>
-<tr id="en-us-ryan-low" class="alt">
+<tr id="en-us-ryan-low" class="">
 <dialog id="dialog-en-us-ryan-low"><form>
 <h1>Model card for Ryan (low)</h1>
 <ul>
@@ -1534,7 +1559,7 @@ <h2>Training</h2>
 <td> <audio id="audio-en-us-ryan-low" preload="none" controls src="samples/en-us-ryan-low/speaker_0/sample.mp3"></audio> <details><summary>sample text</summary><p> A rainbow is a meteorological phenomenon that is caused by reflection, refraction and dispersion of light in water droplets resulting in a spectrum of light appearing in the sky. It takes the form of a multi-colored circular arc. Rainbows caused by sunlight always appear in the section of sky directly opposite the Sun. </p></details> </td>
 <td></td>
 </tr>
-<tr id="en-us-ryan-medium" class="">
+<tr id="en-us-ryan-medium" class="alt">
 <dialog id="dialog-en-us-ryan-medium"><form>
 <h1>Model card for Ryan (medium)</h1>
 <ul>
@@ -2599,6 +2624,30 @@ <h2>Training</h2>
 <td> <audio id="audio-zh-cn-huayan-x-low" preload="none" controls src="samples/zh-cn-huayan-x-low/speaker_0/sample.mp3"></audio> <details><summary>sample text</summary><p> 彩虹，又稱天弓、天虹、絳等，簡稱虹，是氣象中的一種光學現象，當太陽 光照射到半空中的水滴，光線被折射及反射，在天空上形成拱形的七彩光譜，由外 圈至内圈呈紅、橙、黃、綠、蓝、靛蓝、堇紫七种颜色（霓虹則相反）。事實 上彩虹有无数種顏色，比如，在紅色和橙色之間還有許多種細微差別的顏色，根據 不同的文化背景被解讀爲3-9種不等，通常只用六七種顏色作為區別。國際LGBT 聯盟的彩虹旗为六色：紅橙黃綠藍紫。紅橙黃綠藍靛紫的七色說，就是在六色基礎 上將紫色分出偏藍色的靛。傳統中國文化說的七色是：赤橙黃綠青藍紫，青色 就是偏藍的綠色。要是把橙色也分爲偏紅、偏黃的兩種就是九色。三色說有：紅綠 藍，就是光學三原色，所有顏色的光都是這三種顏色混合出來的，和亚里士多 德紅、綠、紫三色說，就是兩頭加中間。 </p></details> </td>
 <td></td>
 </tr>
+<tr id="zh_CN-huayan-medium" class="alt">
+<dialog id="dialog-zh_CN-huayan-medium"><form>
+<h1>Model card for huayan (medium)</h1>
+<ul>
+<li>Language: zh-cn (Chinese)</li>
+<li>Speakers: 1</li>
+<li>Quality: medium</li>
+<li>Samplerate: 22,050Hz</li>
+</ul>
+<h2>Dataset</h2>
+<ul>
+<li>URL: https://github.com/PlayVoice/HuaYan_TTS</li>
+<li>License: Unknown</li>
+</ul>
+<h2>Training</h2>
+<p>Finetuned from U.S. English lessac voice (medium quality).</p>
+<div><button value="cancel" formmethod="dialog">Close</button></div>
+</form></dialog>
+<td> <a title="Download zh_CN-huayan-medium" href="https://github.com/rhasspy/piper/releases/download/v0.0.2/voice-zh_CN-huayan-medium.tar.gz">zh_CN-huayan-medium</a> </td>
+<td> <a href="javascript:q('#dialog-zh_CN-huayan-medium').showModal()">View</a> </td>
+<td> medium </td>
+<td> <audio id="audio-zh_CN-huayan-medium" preload="none" controls src="samples/zh_CN-huayan-medium/speaker_0/sample.mp3"></audio> <details><summary>sample text</summary><p> 彩虹，又稱天弓、天虹、絳等，簡稱虹，是氣象中的一種光學現象，當太陽 光照射到半空中的水滴，光線被折射及反射，在天空上形成拱形的七彩光譜，由外 圈至内圈呈紅、橙、黃、綠、蓝、靛蓝、堇紫七种颜色（霓虹則相反）。事實 上彩虹有无数種顏色，比如，在紅色和橙色之間還有許多種細微差別的顏色，根據 不同的文化背景被解讀爲3-9種不等，通常只用六七種顏色作為區別。國際LGBT 聯盟的彩虹旗为六色：紅橙黃綠藍紫。紅橙黃綠藍靛紫的七色說，就是在六色基礎 上將紫色分出偏藍色的靛。傳統中國文化說的七色是：赤橙黃綠青藍紫，青色 就是偏藍的綠色。要是把橙色也分爲偏紅、偏黃的兩種就是九色。三色說有：紅綠 藍，就是光學三原色，所有顏色的光都是這三種顏色混合出來的，和亚里士多 德紅、綠、紫三色說，就是兩頭加中間。 </p></details> </td>
+<td></td>
+</tr>
 </table>
 
     <hr />

samples/zh_CN-huayan-medium/speaker_0/sample.txt

@@ -0,0 +1 @@
+彩虹，又稱天弓、天虹、絳等，簡稱虹，是氣象中的一種光學現象，當太陽 光照射到半空中的水滴，光線被折射及反射，在天空上形成拱形的七彩光譜，由外 圈至内圈呈紅、橙、黃、綠、蓝、靛蓝、堇紫七种颜色（霓虹則相反）。事實 上彩虹有无数種顏色，比如，在紅色和橙色之間還有許多種細微差別的顏色，根據 不同的文化背景被解讀爲3-9種不等，通常只用六七種顏色作為區別。國際LGBT 聯盟的彩虹旗为六色：紅橙黃綠藍紫。紅橙黃綠藍靛紫的七色說，就是在六色基礎 上將紫色分出偏藍色的靛。傳統中國文化說的七色是：赤橙黃綠青藍紫，青色 就是偏藍的綠色。要是把橙色也分爲偏紅、偏黃的兩種就是九色。三色說有：紅綠 藍，就是光學三原色，所有顏色的光都是這三種顏色混合出來的，和亚里士多 德紅、綠、紫三色說，就是兩頭加中間。

zh_CN/huayan/medium/MODEL_CARD

@@ -0,0 +1,15 @@
+# Model card for huayan (medium)
+
+* Language: zh-cn (Chinese)
+* Speakers: 1
+* Quality: medium
+* Samplerate: 22,050Hz
+
+## Dataset
+
+* URL: https://github.com/PlayVoice/HuaYan_TTS
+* License: Unknown
+
+## Training
+
+Finetuned from U.S. English lessac voice (medium quality).