Skip to content

docs: README rewrite + competitive positioning (T2-2.5, 错位竞争) - #129

Merged
TeFuirnever merged 2 commits into
mainfrom
docs/rewrite-readme-t2-2-5
Jul 31, 2026
Merged

docs: README rewrite + competitive positioning (T2-2.5, 错位竞争)#129
TeFuirnever merged 2 commits into
mainfrom
docs/rewrite-readme-t2-2-5

Conversation

@TeFuirnever

Copy link
Copy Markdown
Owner

Summary

执行 strategic-plan §T2-2.5 的 README 重写 + 新增竞品定位文档。定位决策为错位竞争(2026-07-31 与项目维护者确认):Murmur 不与系统听写正面竞争实时性,聚焦文件转录 + AI 后处理 + 完全本地。

改动

1. README.md 重写(+55/-37)

  • Hero 区:价值主张强化为「说话 → 文字 → AI 润色 → 自动粘贴」,加醒目 GIF 占位 + 维护者录制 TODO
  • 定位说明:显式声明不与系统听写正面竞争实时性,聚焦三个差异化
  • 竞品对比表:行序改为差异化优先(文件转录、AI 后处理、自定义 Prompt、11+ 模型在前,中文精度在后),加共存说明承认系统听写实时优势
  • 特性网格:文件转录提到第一行(原来被埋)
  • 路线图链接:指向 follow-ups.md + CHANGELOG(活文档),不再指向已冻结的 strategic-plan-gap-analysis.md
  • 英文版:对齐中文版结构(原来更短,缺定位说明/GIF 占位/差异化优先的对比表)

2. docs/competitive-positioning.md(新建,+114)

回答 strategic-plan 评审识别的核心盲区:"用户为什么不用系统自带语音输入?" 包含:

  • 正面竞争的死局(架构 + 生态不可逆)
  • 错位竞争的胜算(系统听写结构性盲区)
  • 核心用户画像(5 类,明确放弃实时打字用户)
  • 竞品矩阵与护城河排序
  • 战略风险与缓解(Apple Intelligence 3-6 月威胁)
  • 成功指标(按错位定位调整:AI 激活率 + 文件转录使用率,非实时延迟)
  • 决策记录(待 3-6 月验证后转 ADR)

定位决策依据

  • strategic-plan 三轮评审共识:"17 项行动对 1 人团队不现实",需聚焦
  • 架构现实:非流式 ASR,全栈重构成本高
  • 系统听写生态优势不可逆(OS 一等公民 + Apple Intelligence)

验证

  • pnpm format:check
  • 纯文档改动,无代码变更

后续

  • README 的演示 GIF 需维护者本人录制(10 秒:说话→文字→AI润色→粘贴),我无法代劳
  • 如果错位定位在 3-6 月内验证有效(AI 激活率 + 文件转录使用率达标),competitive-positioning.md 应转为正式 ADR

woshiguanxiaoliang added 2 commits July 31, 2026 13:15
Apply the strategic-plan §T2-2.5 README rewrite with '错位竞争'
(dislocated competition) positioning decided with user:

- Hero: strengthen value prop to 'speak → text → AI polish → paste',
  add prominent GIF placeholder + maintainer recording TODO
- Positioning note: explicitly state Murmur doesn't compete head-on
  with system dictation on real-time latency; focuses on file
  transcription + AI post-processing + fully local + custom models.
  Real-time streaming acknowledged as roadmap item.
- Comparison table: reorganize rows to lead with differentiators
  (file transcription, AI post-processing, custom prompts, 11+ models)
  instead of generic 'Chinese accuracy'. Add a coexistence note
  acknowledging system dictation's real-time advantage.
- Feature grid: promote file transcription to first row (was buried),
  reflect batch + AI polish capabilities.
- Roadmap link: point to follow-ups.md + CHANGELOG (live sources)
  instead of strategic-plan-gap-analysis.md (frozen historical
  snapshot, per 2026-07-31 freeze notice).
- English section: align structure with Chinese (was shorter and
  missing positioning note, GIF placeholder, and differentiator-first
  comparison table).
Answer the core strategic blind spot identified by strategic-plan
reviewers: 'why would users not use system dictation?'

Decision: dislocated competition (错位竞争) — Murmur does NOT compete
head-on with macOS/Windows system dictation on real-time latency.
Instead it focuses on three things system dictation can't do:

1. File transcription (system dictation is an input method, not a
   file processor)
2. AI post-processing (remove filler, fix errors, format as meeting
   notes / Xiaohongshu / Zhihu — system dictation only does raw
   transcription)
3. Fully local + customizable models (Ollama / 11+ providers —
   system dictation uses Apple private models, non-replaceable)

Key insight: these aren't things system dictation does poorly, but
things it structurally won't do (product category difference, not
feature gap).

Includes:
- Why head-on competition is a dead end (architecture + ecosystem)
- Core user personas (writers, meeting note-takers, privacy users,
  content creators, devs — NOT real-time typists)
- Competitor matrix (direct: Whisper Desktop/iFlytek/Feishu;
  indirect: system dictation — coexist, not replace)
- Moat ranking (AI post-processing strongest; Chinese accuracy
  needs FunASR caveat)
- Risk mitigation (Apple Intelligence 3-6mo threat, etc.)
- Success metrics re-aligned to positioning (AI activation rate +
  file transcription usage, NOT real-time latency)
- Decision record (should convert to ADR after 3-6mo validation)
@TeFuirnever
TeFuirnever merged commit 51b38b5 into main Jul 31, 2026
1 check passed
@TeFuirnever
TeFuirnever deleted the docs/rewrite-readme-t2-2-5 branch July 31, 2026 05:32
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant