NetEase Tianyin
Lyrics, arrangement and singing in three stages; fast Chinese drafts, finish them yourself.
Model profile
- Vendor
- 网易（网易云音乐）
- Form
- 网页服务
- Input
- 文本 · 歌词 · 参考音频
- Output
- 音频 · 分轨
- Licence
- 词曲编唱的产出归属、商用与发行按网易天音与网易云音乐的服务条款；活动规则与付费档会变
- Verified
- 2026-10-31
Related
Lyrics, arrangement and singing in three stages
Tianyin splits songwriting into three work areas: lyric assistance proposes lines from a theme and a syllable pattern, arrangement builds a backing from a style, and singing synthesises words and melody into a voice. You can also generate a whole draft from a theme and return to the three areas to revise. It targets people without theory training, so chords and arrangement terms stay hidden, at the cost of fine control. Changing a drum fill or one harmony usually means exporting to a DAW.
From a theme to a draft
Input can be a theme line, a short note, a few lyric lines or a reference clip; output is stereo with vocals, and some versions and tiers export stems. After generation you can rewrite words, adjust melody, swap voices and change reverb. Its strength is turning an idea into a listenable demo, not turning a demo into a release master. Arrangement depth, drum groove and vocal breath still need your work.
| Stage | What you can change | Check before delivery |
|---|---|---|
| Lyrics | Theme, syllable pattern, sections | Tones and stresses |
| Arrangement | Style, instruments, tempo | Low end and drum density |
| Singing | Voice, mood, reverb | Diction and breath |
| Export | Mix, some stems | Format and sample rate |
Spoken naturalness in Chinese
Tianyin's Mandarin diction is natural on common pop lines; light tones, erhua and sentence-final particles sit close to speech. Tones still get pulled by the melody, and a run of high level tones can turn into chanting. Put stressed syllables on strong beats and long notes on open vowels and the result steadies. Guofeng, folk and short-video beds are its common uses, and the platform's word bank and templates lean that way.
Templates and export
There are plenty of arrangement templates, which is fast but makes it easy to hear the same backing with a different tune. Stem export and track merging depend on version and tier, and the project file may not come with you; moving to another platform or into a DAW often means realigning. Voices are selectable but consistency across runs is modest, so two stitched lines can sound like two singers. One-click release saves steps, but credits, AI disclosure and rights data are still yours to fill in.
Questions before release
Confirm: AI disclosure, ownership of words and music, whether streaming release is allowed, the commercial scope of stems, and any extra terms attached to campaigns. The current NetEase Tianyin and NetEase Cloud Music terms are the authority. Alternatives: MiniMax Music for stronger Mandarin full songs and an API; TME Tianqin for stems and platform release; YuE or DiffRhythm for local control.
Listen in the library
All tracks from midicn-lib — inline badge is the license tier (C1 commercial / C2 non-commercial / C3 study only).
Along the same line
Sources
- 网易天音官方产品页与网易云音乐创作入口（作词、编曲、演唱与导出）。
- 网易天音公开更新说明中的一键写歌、词曲协同与发行入口。
- 网易云音乐服务条款与 AI 创作活动规则（产出使用与商用）。