下面整理了一组“在线音频处理”方向的在线工具生成型 AI 提示词（Prompts）。每条 Prompt 都要求 AI 产出完整可运行项目：包含源码、清晰文件结构、可复现的运行命令、部署说明，以及不少于 5 条测试用例或 QA checklist，适合做成可交付的在线小工具。
在线批量音频转码工具（多格式输入输出）
面向需要把 WAV/MP3/AAC/FLAC/OGG 等批量转为指定格式与参数的场景，支持队列与打包下载。
英文 Prompt:
You are to generate a complete, runnable web app: an online batch audio transcoder.
Requirements:
- Category: audio processing only.
- Tech stack: Next.js (App Router) + Type + Tailwind CSS.
- Use ffmpeg.wasm (WebAssembly) in a Web Worker for heavy processing; UI must stay responsive.
- Features: multi-file upload (drag & drop), per-file progress, preset profiles (mp3 320k, mp3 128k, aac m4a, opus, flac), custom settings (bitrate, sample rate, channels), queue management, cancel, retry.
- Output: download each file + "Download all" as a zip.
- Security: do everything client-side; no server upload.
Deliverables (must include all):
1) Project file tree.
2) Full source code for every file.
3) Exact commands to run locally.
4) Deployment guide (Vercel) and browser compatibility notes.
5) A test plan: at least 5 concrete test cases or a QA checklist (include edge cases like large files, unsupported codecs, cancel mid-job, multiple parallel jobs).
Also include a short README explaining the architecture (UI thread vs worker, ffmpeg loading, memory limits).
中文释义: 让 AI 直接生成一个“可部署的在线音频转码器”项目，支持批量处理、预设与自定义参数、进度与打包下载，并给出运行与部署步骤及完整测试清单。
在线响度归一工具（EBU R128 / LUFS）
适合播客、课程、短视频配音等需要把多段音频统一响度的场景，避免忽大忽小。
英文 Prompt:
Build a complete online audio loudness normalizer web app.
Constraints:
- Client-side only (no server storage).
- Next.js + Type + Tailwind.
- Use ffmpeg.wasm in a Web Worker.
Features:
- Upload one or multiple audio files.
- Analyze loudness and show metrics (target LUFS, measured integrated loudness, true peak).
- Normalize to a user-selected target (e.g., -16 LUFS, -14 LUFS) using ffmpeg filters.
- Allow output format selection (mp3, m4a, wav).
- Show per-file report and allow exporting the report as JSON.
Deliverables:
- File tree + complete source code.
- Local run commands.
- Deployment instructions.
- At least 5 test cases/QA checks (include clipping prevention, true-peak limit, mixed sample rates, batch consistency, invalid input handling).
Important: include a clear disclaimer about limitations of client-side ffmpeg and large-file memory usage.
中文释义: 让 AI 生成一个在线“响度分析+归一化”的完整项目，实现批量处理、报告导出与可部署交付，并提供不少于 5 条测试/自检项。
在线音频降噪工具（语音优先，不做“生成音效”）
针对录音底噪、风噪、空调声等，输出更干净的人声文件；强调处理与增强，不涉及生成式音频。
英文 Prompt:
Create a full web app: an online speech-oriented audio denoiser.
Constraints:
- Do NOT generate new audio content; only noise reduction/enhancement on existing recordings.
- Next.js + Type + Tailwind.
- Prefer an on-device approach:
- Option A: RNNoise WASM (recommended) for speech denoising.
- Option B: ffmpeg.wasm filters (if RNNoise is too complex).
- Must run fully in-browser.
Features:
- Upload audio, preview before/after with an A/B toggle.
- Adjustable strength (light/medium/strong) and preserve-voice option.
- Export as wav or mp3.
- Show processing time and a simple waveform preview (wavesurfer.js).
Deliverables:
- File tree, complete code, run commands, deploy steps.
- Provide a QA checklist with at least 5 items (include artifacts, over-suppression, different codecs, long recordings, mobile limitations).
- Include a short section explaining how the denoising model/filter works and its limitations.
中文释义: 生成一个“在线降噪”工具项目，重点是对现有音频做降噪/增强并可预览对比，包含完整源码、部署与 QA 清单，不涉及任何生成式音频。
在线波形剪辑与分段导出工具（Trim / Split）
适合把长录音快速剪掉空白、按时间点切段并批量导出。
英文 Prompt:
Generate a complete online waveform editor for trimming and splitting audio.
Stack:
- Next.js + Type + Tailwind.
- wavesurfer.js for waveform and region selection.
- Use ffmpeg.wasm in a Web Worker to cut segments accurately.
Features:
- Upload an audio file and display waveform.
- Create multiple regions; support snap-to-silence (simple silence detection) to help quick cuts.
- Export: each region as a separate file, plus a zip download.
- Preserve original encoding when possible; otherwise re-encode with selectable settings.
Deliverables:
- Full code, file tree, commands.
- Deployment guide.
- At least 5 test cases/QA checklist items (accuracy of cut points, region overlaps, silence detection correctness, large files, export naming rules).
Also include accessibility notes (keyboard region selection, ARIA labels).
中文释义: 让 AI 产出一个可部署的在线波形剪辑/切段工具，支持区域选择、静音辅助切割、批量导出与压缩包下载，并给出完整测试清单。
在线音频元数据编辑器（ID3/标签/封面字段不上传）
用于整理曲名、专辑、作者、年份、流派等信息；仅修改现有文件的标签，不把文件上传到服务端。
英文 Prompt:
Build a complete in-browser audio data editor.
Constraints:
- Client-side only.
- Next.js + Type + Tailwind.
- Support common formats: mp3 (ID3v2), m4a, flac, ogg where feasible.
Features:
- Batch select files and edit fields in a table ( /artist/album/year/genre/track).
- Validation (required fields, numeric ranges).
- Write back tags and let users download updated files.
- Provide a "diff" preview showing what data will change.
Deliverables:
- File tree, complete source code, run commands.
- Deployment instructions.
- QA checklist with at least 5 items (tag reading, writing integrity, non-ASCII characters, batch apply/undo, unsupported formats).
Include notes on libraries chosen for data parsing/writing and their limitations.
中文释义: 生成一个在线“音频标签/元数据批量编辑器”项目，包含变更预览与校验、完整源码和可复现部署步骤，并覆盖常见边界情况的 QA 检查项。
在线音频格式与编码信息检查工具（导出诊断报告）
用于排查“为什么无法播放/转码慢/音画不同步（仅音频场景）”等问题，输出可分享的诊断结果。
英文 Prompt:
Create a complete web tool: audio file inspector + diagnostics report.
Stack:
- Next.js + Type .
- Use ffprobe via ffmpeg.wasm (or equivalent data extraction) in a Web Worker.
Features:
- Upload one or multiple audio files.
- Show codec, container, duration, sample rate, channels, bitrate, tags.
- Highlight potential issues (VBR, unusual sample rates, missing duration, corrupted headers).
- Export report as JSON and as a copyable text summary.
Deliverables:
- File tree, full code, run instructions, deploy guide.
- QA checklist with at least 5 items (corrupt file handling, multiple formats, performance, report correctness, copy/export behavior).
中文释义: 让 AI 生成一个在线“音频信息检查+诊断报告导出”工具项目，帮助快速查看编码参数与潜在问题，并能导出报告用于沟通排查。
在线批量音频重采样与声道转换工具（Sample Rate / Mono/Stereo）
面向语音识别、客服录音、音频素材库等对采样率与声道有统一要求的场景。
英文 Prompt:
Generate a complete web app for batch resampling and channel conversion.
Requirements:
- Client-side processing with ffmpeg.wasm in a Web Worker.
- UI: batch upload, choose target sample rate (8k/16k/44.1k/48k), choose channels (mono/stereo), choose output format (wav/mp3).
- Show before/after file stats.
- Provide presets for common ASR pipelines.
Deliverables:
- Full project code + file tree.
- Local run commands.
- Deploy instructions.
- At least 5 QA checks (audio quality sanity, exact sample rate verification, mono conversion correctness, batching, cancel/retry).
Add an explanation of how to verify sample rate/channels using the tool itself.
中文释义: 产出一个在线“批量重采样/声道转换”工具项目，强调可运行交付与自检方法，适合 ASR/统一规格的音频处理工作流。
在线音频批处理流水线工具（可配置步骤并导出配置）
把“转码 + 响度归一 + 重采样 + 命名规则”等组合成可复用的处理流水线，适合重复任务。
英文 Prompt:
Build a complete online audio batch pipeline builder.
Constraints:
- Audio processing only; no image generation.
- Next.js + Type + Tailwind.
- Use ffmpeg.wasm in a Web Worker.
Features:
- Users can define a pipeline with steps: transcode, normalize loudness, resample, trim start/end, rename with template.
- A visual step list UI with drag-and-drop reorder.
- Save/load pipeline config as JSON.
- Batch run on multiple files with progress and error reporting.
- Export outputs + processing log.
Deliverables:
- Complete code, file tree, local run commands.
- Deployment guide.
- QA checklist >= 5 items (step ordering correctness, config import/export, error handling, performance/memory limits, deterministic output naming).
Include a brief architecture section: step graph, worker messaging protocol, and how you keep the UI responsive.
中文释义: 生成一个可部署的“音频批处理流水线构建器”，能把多步音频处理组合成可复用配置并批量执行，输出日志与结果文件，并提供完善 QA 清单。
