1. V7 完整产品流程
1. V7 full product flow
打开 V7，点击完整流程，然后检查歌曲、分轨、声线版本、QA 门禁、MP3 下载、STEMS ZIP 和 QA 报告。
Open V7, run the full flow, then verify song, stems, voice version, QA gate, MP3 download, STEMS ZIP, and QA report.
OpenUSAGE / 使用说明 / Інструкція
从注册、免费试用、生成、分轨、声线、质检到下载，按这里走。
Start with V7 for the product-grade flow. Use Create, Upload, Voices, Studio, Library, and Pricing when you need the single-purpose tools. The local app runs on 13910/13911 and keeps real production execution blocked until credentials, storage, database, queue, and voice data are ready.
Local app13910 / 13911
V7 smokeMP3 / ZIP / QA
ExportsWAV / MP3 / STEMS
2. 生成完整歌曲
2. Create a complete song
输入一句话，选择语言、风格、情绪、BPM、调性、时长和人声类型，然后生成。
Enter a prompt, choose language, genre, mood, BPM, key, duration, and vocal type, then generate.
Open3. 选择现代音乐风格
3. Choose modern styles
支持 Pop、R&B、HipHop、Trap、Drill、EDM、House、Techno、LoFi、Jazz、Rock、Metal、国风、K-pop、J-pop、City Pop、影视配乐、游戏配乐和趋势组合。
Use Pop, R&B, HipHop, Trap, Drill, EDM, House, Techno, LoFi, Jazz, Rock, Metal, Guofeng, K-pop, J-pop, City Pop, cinematic, game score, and trend combos.
Open4. 上传音频并分轨
4. Upload and split audio
先确认版权，再上传 MP3/WAV 或加载示例音频，分析 BPM/Key，然后分离 vocals、instrumental、drums、bass、other。
Confirm rights, upload MP3/WAV or load demo audio, analyze BPM/key, then split vocals, instrumental, drums, bass, and other.
Open5. 训练并使用私有声线
5. Train and use a private voice
上传已授权样本，创建私有唱歌转换声线。不要把 TTS 朗读当成唱歌。
Upload an authorized sample and create a private singing-conversion voice. Do not treat TTS reading as singing.
Open6. 在 AI Studio 混音
6. Mix in AI Studio
查看波形、播放混音、静音/独奏分轨、调音量，并导出 WAV 或 stems ZIP。
Inspect waveforms, play the mix, mute/solo stems, adjust volumes, and export WAV or stems ZIP.
Open7. 定价和付费流程
7. Pricing and payment flow
先免费使用一次。付费套餐跳转 TJ 支付入口；网站本身不直接执行资金动作。
Use one free generation first. Paid plans route to the TJ payment hub; the site does not move money by itself.
Open8. 作品库和证据
8. Library and evidence
可回放、发送到 Studio、下载资产，并保留歌曲、分轨、声线、导出和 QA 报告证据。
Replay, send work to Studio, download assets, and keep generated songs, stems, voices, exports, and QA reports visible.
Open9. V6 开源引擎研究
9. V6 open-engine research
用 V6 查看开源引擎候选和产品边界，再接真实运行时。
Use V6 to review the open-source engine shortlist and product boundary before connecting real runtimes.
OpenACCEPTANCE CHECKS
交付前按这几条验收。
V7 route: http://127.0.0.1:13910/v7
Guide route: http://127.0.0.1:13910/guide
Backend QA route: http://127.0.0.1:13911/api/system/qa-report
This project uses 13910/13911 and does not occupy localhost:8xxx.
Generated downloads must be real files, not 1 KB placeholders.
Perfect Longge singing needs dry singing data, verified runtime output, and human listening approval.