voice-input

Before you start: read this whole install prompt, tell me in a few lines what this skill does and what installing it involves, and wait for my go-ahead before you do anything else. Prerequisites: none for the install itself: no account, no API key, no paid service, and nothing else to install first. For using the skill afterwards as described, you need a fal account and API key (environment variable FAL_KEY) for its speech-to-text endpoint — the README lists fal among NPCTalk's paid services, while SKILL.md itself names no price for the Nemotron ASR model — and a game or mod integration that captures the microphone and routes what was heard to the target NPC; Whisper (any host) is the alternative SKILL.md names instead of fal. Never put the key in a file. Install "voice-input" for me. It is a single SKILL.md playbook for letting the player talk to NPCs out loud: an always-on open mic with voice activity detection against the room's noise floor, push-to-talk and typing as fallbacks, a pause so the NPC's own voice coming out of the speakers is not transcribed as the player's, drops for coughs, clicks and filler words, fast speech-to-text on fal in about 0.4 s with the audio sent inline, and routing of what was heard to the right NPC. Repository: https://github.com/christos-antono/NPCTalk (the skill is the .agents/skills/voice-input/ folder at the pinned commit: SKILL.md only; the identical .claude/skills/ copy and the root skills/ copy are duplicates, not the source of this listing, and the rest of the NPCTalk repo is not installed) Pinned version: install the code at commit 59635ceb8ee010599c3caf9d9d1ee1aa114af91d exactly. Do not substitute a newer version. If this commit is unreachable, stop and tell me. 1. Clone https://github.com/christos-antono/NPCTalk into a temporary folder, check out commit 59635ceb8ee010599c3caf9d9d1ee1aa114af91d, and copy the .agents/skills/voice-input/ folder from that checkout. Then compare every file in the folder with the fingerprints recorded for this listing on 2026-10-19: [FINGERPRINTS] - SKILL.md: SHA-256 67fc1282cae170109ec2396b28035be14609a5904accf61f697743ba3cd4fe1f If a file does not match, it changed since the listing was prepared: do not install it yet, tell me what you found and let me decide. 2. Read SKILL.md in full and tell me in plain words: what the skill does when used, which services it connects to, whether it reads or sends any personal information or keys, and anything that seems unusual. Stop on any red flag and tell me before going further. Confirm these points from the files: - The skill folder holds one file only: SKILL.md (3,217 bytes). There is no code, no script, no reference file and no dependency list in the folder: the skill is instructions for capturing the player's voice, detecting speech against the room's noise, transcribing it and routing it to the right NPC. - Nothing is sent anywhere during installation itself; the only download at install time is the skill folder from GitHub at the pinned commit. The file names no telemetry and no remote service other than fal speech-to-text at use time, described next. - At use time, and only then, the design sends the player's voice audio to fal: each detected utterance is a WAV sent for transcription. In the GTA V example code that SKILL.md points to (examples/gta-v/mod/SpeechToText.cs, not part of this skill and not installed), the clip goes in a POST to https://fal.run/nvidia/nemotron-asr-multilingual/asr with header "Authorization: Key $FAL_KEY", the body carries the WAV inline as a data URI ("audio_url": "data:audio/wav;base64,...", so there is no separate upload step), and the endpoint is the fal model nvidia/nemotron-asr-multilingual/asr named by SKILL.md, returning "output" or "text" in about 0.4 s for a short sentence (0.38-0.80 s logged). That is a disclosure, not a step to run during this install. Note what that means: the open mic hears everything in the room — a call, music, someone else talking — so audio that is not paused or muted can be captured and sent for transcription in the same way. - Capture and detection, in the open-mic design SKILL.md describes (examples/gta-v/mod/OpenMic.cs in the example, not installed): 16 kHz mono capture in 30 ms blocks; the noise floor is the quietest moments of the last 1.5 s; an utterance starts when the level stays SensitivityDb (12 dB) above that floor for 150 ms and ends after SilenceMs (700 ms) below it; the floor adapts to the room and to the game's own sound; 300 ms from before the start is kept so the first syllable is not cut; anything shorter than 450 ms (coughs, clicks) and anything over 12 s without a pause (music, a loud game) is dropped. - Pause while an NPC talks, and for 600 ms after: their voice comes out of the speakers and would be transcribed as the player's. Also pause while push-to-talk is held, while typing, and while a reply is being written; headphones make this much easier. After transcribing, drop non-speech: one-word fillers ("uh", "um", "hmm", "yeah", "okay", "thanks"...) and empty transcripts are ignored. - Three ways in, all ending in the same exchange: open mic (just talk), push-to-talk (hold N) and type (B); build push-to-talk first, then the open mic. Who it goes to is whoever the target lookup returns at that moment (skill make-npcs-talk, hook 1, not part of this skill and not installed): within 5 m, or looked at within 12 m; the last partner sticks while close. If nobody is there, it is logged ("open mic -> nobody near") and shown on screen ("Listening... (to nobody near you)"); the player also gets a mute key (M) and the mic state on screen. SKILL.md also describes testing without the game with npcvoice-check vad in the GTA example (the detector run on real voice clips mixed over quiet and loud noise) and debugging cases ("nobody answers", the NPC answers the room, it triggers on the game, first word missing); do not run any of that during this install. - Without fal, SKILL.md says Whisper (any host) is the alternative; whatever you use, measure it per exchange. If I later choose a different host myself, the audio would go to that host instead — that is not the default described here and is not part of this install. 3. When I say "yes, install", install it as a skill: copy the whole voice-input folder into the agent's skills directory, in a folder named "voice-input", keeping its file (SKILL.md). Do not use the repo's plugin, marketplace or npx routes: those install the whole NPCTalk plugin or all of its skills, not this single skill alone. 4. Verify with no network calls: frontmatter valid, file in place. 5. Report what was installed, where, and what I still need to do myself (for example: get a fal API key myself and set FAL_KEY as an environment variable, never in a file, only if and when I want to use fal speech-to-text; build or provide the game-side microphone capture, voice activity detection and target lookup myself; provide any game, any microphone and any speech myself; using the skill is a separate step that only starts when I ask for it, and the use-time send of voice audio to fal disclosed above happens only then, on utterances I run). GitHub is optional: if I have a GitHub account or the gh CLI, you may use it; otherwise public access is fine. Never require it unless it's in the prerequisites above. Rules: don't touch anything outside the temp folder and the install target. If anything looks off, stop and ask me.