Spoken Magpie Ja: Japanese Speech Synthesis Dataset for Instruction Tuning

Spoken Magpie Ja is a Japanese speech synthesis dataset created for instruction tuning of language models. It was generated using the CosyVoice2 TTS system on the llm-jp/magpie-sft-v1.0 text corpus to produce a commercially usable dataset. The dataset was last updated on 2026-01-12.

Use Cases

• Training Japanese text-to-speech models based on the described instruction-response pairs.
• Fine-tuning voice-enabled language models for instruction-following tasks based on the provided audio-text pairs.
• Developing conversational AI agents with spoken Japanese output capabilities based on the audio responses.
• Creating educational tools for Japanese language learning based on the spoken instruction-response format.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.

Provenance

Source

Atotti on Hugging Face.

Collection Method

Generated by applying CosyVoice2 TTS to the llm-jp/magpie-sft-v1.0 text corpus.

Freshness

Last updated 2026-01-12 05:29:21; freshness should be verified.