Inspect Classical Chinese poetry through reconstructed IPA
A Qieyun-based reconstruction, not a definitive historical recording
Your IPA transcription will appear here...
Explore reconstructed readings for familiar poems
Middle Chinese (中古漢語) is the reconstructed phonological stage commonly used to discuss Sui, Tang, and early Song rhyme traditions.
The sounds behind Tang poetry differed dramatically from modern Mandarin. This tool exposes one reconstruction as inspectable IPA and labels audio as an approximation.
This system keeps reconstructed IPA as the source of truth and treats speech synthesis as an optional listening aid:
Converts Chinese characters to Middle Chinese IPA using the ToMiddleChinese library, based on unt's scholarly reconstruction of the Qieyun system (601 CE). This exposes the distinctions represented by that reconstruction in IPA form.
Maps Middle Chinese IPA to eSpeak-NG phoneme notation through custom mappings. Features like uvular stops, retroflex consonants, and pharyngealization are approximated for synthesis.
Uses ElevenLabs Speech-to-Speech API with the multilingual model to smooth robotic eSpeak-NG output. This is optional voice enhancement, not scholarly evidence.
Middle Chinese reconstructions model a rich consonant inventory and syllable-final contrasts reflected in rhyme-book categories. The Qieyun tradition records categories that modern reconstructions interpret as IPA values.
The IPA comes from one reconstruction system through ToMiddleChinese. It is useful for exploration and comparison, but it is not a definitive recording of how any individual poet spoke.
The Qieyun tradition records phonological categories rather than microphone-like sound. Later rhyme books such as Guangyun preserve and expand that tradition, but any IPA output is still a modern reconstruction.
Audio is generated by mapping reconstructed IPA into modern speech synthesis tools, so it should be heard as an approximation of the notation rather than independent historical evidence.