CLAUDE: VOICE & MOBILE
5 SRC
Claude: Voice & Mobile
Voice input captures rough thought before polished compression, preserving uncertainty, emphasis, and unfinished lines of thought. Raw meeting transcripts and dictated planning notes outperform short summaries as Claude source material because the model fills transcription gaps and works from messy input. Monologue (from Every) or WhisperFlow pipes speech into any focused app; a gooseneck microphone is recommended. Mac Mini with Telegram integration enables mobile Claude Code access—send commands like /ce:plan fix the timeout issue from anywhere, with tmux sessions surviving bad WiFi. Voice modeling is a reusable skill workflow: analyze several weeks of substantive, audience-separated Slack and email to infer what the person challenges, notices, and ignores; require example-supported patterns; show drafts before saving; update from gaps between draft and actual sends. Claude Sonnet 4.5 replicates individual writing voices with 94% accuracy in blind tests across literary, essay, and professional styles, enabling executives to delegate communication while maintaining authentic voice. Build reusable skills (e.g., $tibo-voice) from these patterns. Voice's value is fidelity, not speed: the agent receives unedited, messy thinking, giving it richer context than polished typed prompts.
Insights
Voice tools: Monologue (@usemonologue) or WhisperFlow: Monologue (from Every) pipes speech into whatever app is focused. Voice-to-LLM works because Claude understands context and fills in transcription gaps — you can mumble, trail off, restart sentences. Gooseneck microphone recommended. (from every claude code hack mvanhorn)
Mac Mini + Telegram = mobile Claude Code: Telegram integration lets you send commands from your phone.
/ce:plan fix the timeout issuefrom dinner → plan waiting in Zed at home. tmux sessions on the Mini survive bad WiFi — reconnect and pick up where you left off. (from every claude code hack mvanhorn)Claude Sonnet 4.5 can replicate individual writing voices with 94% accuracy in blind tests, working across literary (Hemingway), essay (Paul Graham), and professional (CEO email) styles — executives can delegate communication while maintaining authentic voice (from clone writing voice claude)
Ask it to identify patterns: what they push back on, what piques their interest, what they ignore, and how they communicate with their team — then build a reusable skill (e.g. $tibo-voice) from those patterns — relates to Voice (from codex voice skill from slack mentor)
Voice as Input Medium
- Voice input captures rough thought before compression into polished prose. Raw meeting transcripts or dictated planning notes outperform short summaries as source material because they preserve uncertainty, emphasis, and unfinished lines of thought. (from codex work os patterns)
Voice as high-bandwidth context
- Voice input's value is not speed but fidelity: the agent receives the unedited, messy version of your thinking. Combining Wispr Flow (system-wide dictation) with Codex's built-in voice, and piping call transcripts via Granola, gives the model richer context than polished typed prompts ever would. (from codex maxxing jason liu)
Voices
8 contributors
jason
@jxnlco
hype @openai
Nick Spisak
@NickSpisak_
| AI Transformation Engineer | Seven Figure E-Commerce Business Owner
Nick Khami
@skeptrune
currently doing things at Mintlify, prev. built a search API (trieve acq. YCW24), you should try to fail faster
Samantha Trimble
@strimblez
the other sam at openai
Griffin Hilly
@GriffinHilly
Bond trader by day, pro-humanism/nuclear on my own time. @MadiHilly’s biggest fan
Manthan Gupta
@manthanguptaa
ai research engineer • designing agent runtimes, memory & retrieval systems for autonomous agents • dms open
rLLM
@rllm_project
Enabling AI agents to "learn from experience" @BerkeleySky Try Hive: https://t.co/S9kJjTWgA9
Shiv
@shivsakhuja
Pontificating... / Vibe GTM-ing / Making Claude Code do non-coding things building a team of AI coworkers @ Gooseworks / prev @AthinaAI /@google / @ycombinator