HuggingFace dataset + Chinese notes

Makefile

@@ -15,3 +15,7 @@ yali-lower.exe: wav-syllables setup.bat yali-exe-readme.txt unzipsfx.exe
 	cat unzipsfx.exe yali-lower.zip > yali-lower.exe
 	zip -A yali-lower.exe
 	rm -rf partials yali-lower.zip
+
+metadata.csv: # for HuggingFace
+	echo filename,syllable,tone,tone_description,pitch_adjusted > $@
+	for N in wav-syllables/*.wav; do echo $$N,$$(echo $$N|sed -e s,.*/,, -e s/[1-6].*//),$$(echo $$N|sed -e s,[^1-6],,g),$$(case "$$(echo $$N|sed -e s,[^1-6],,g)" in 1) echo high level;; 2) echo rising;; 3) echo low dipping "(half)";; 4) echo falling;; 5) echo neutral;; 6) echo "neutral (high)";; esac),true; done >> $@

README.md

@@ -1,3 +1,28 @@
+---
+# Metadata for HuggingFace dataset viewer （请垂直下移看汉语解释）
+# (Humans please skip past this.  Why does HuggingFace insist it comes first?...)
+license: gpl-3.0
+task_categories:
+- text-to-speech
+- automatic-speech-recognition
+language:
+- zh
+modality:
+- audio
+size_categories:
+- 1K<n<10K
+tags:
+- mandarin
+- syllables
+- tones
+- speech-synthesis
+- phonetics
+- native-speaker
+---
+
+# Yali-Lower: Isolated Mandarin Syllables with Full Tone Coverage (lower-pitch version)
+## Yali语音：带完整声调的孤立汉语音节录音库
+
 These syllables were recorded in context by
 Cheng Ya Li for the [Gradint](../gradint) program
 in June 2008.  Sound editing was done by Cameron Wong
@@ -5,6 +30,31 @@ in June 2008.  Sound editing was done by Cameron Wong
 speech synthesizer) and Silas S. Brown, using
 Audacity and Praat.
 
+这些音节由程雅丽（Cheng Ya Li，汉字待确认）于2008年6月为
+[Gradint](../gradint)程序录制。音频编辑由黄冠能（Cameron
+Wong，[Ekho](https://github.com/hgneng/ekho)语音合成器的开
+发者）和Silas S. Brown（赛乐思）使用Audacity和Praat完成。
+
+## Technical Notes
+- **Tone 1 and Tone 6**: Slightly pitch-adjusted to
+  compensate for pitch drift across recording session,
+  ensuring consistent tone perception when syllables are
+  synthesized together
+- **Tones 2, 3, 4, 5**: Original pitch preserved from
+  context extraction
+- **Tone 3**: Recordings represent the "half tone 3"
+  commonly used in natural speech
+- **Tone 6**: Higher-pitched neutral tone variant
+  typically used after Tone 3
+
+技术说明
+- **第一声和第六声**：已进行轻微音调调整以补偿录音时段间的音调漂移
+- **第二、三、四、五声**：保留从语境中提取的原始音调
+- **第三声**：录音为自然语音中常用的"半三声"
+- **第六声**：通常在第三声后使用的高音中性调
+
+## Compiling for Gradint
+
 `yali-lower.exe` is a zip file that must be
 compatible with unzip on non-Windows platforms
 as well as being self-extracting.
@@ -30,3 +80,8 @@ These recordings are distributed in the hope that they will be useful,
 but WITHOUT ANY WARRANTY; without even the implied warranty of
 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 General Public License for more details.
+
+您可以根据自由软件基金会发布的 GNU 通用公共许可证（GPL）条款重新分发和/或修改这些录音；许可证版本为第 3 版或任何更新版本。 
+
+这些录音以“按原样”方式提供，不提供任何形式的担保，包括但不
+限于适销性或特定用途适用性的默示担保。