--- name: pp-fish-audio description: "Fish Audio text-to-speech, voice cloning, and transcription from the terminal, with a local render log that tracks every byte and dollar the API never records. Trigger phrases: `render this text to speech with fish audio`, `clone a voice from this clip`, `transcribe this audio with fish audio`, `how much have we spent on fish audio`, `verify this cloned voice`, `use fish-audio`, `run fish-audio`." author: "Jon Gouveia" license: "Apache-2.0" argument-hint: " [args] | install cli|mcp" allowed-tools: "Read Bash" metadata: openclaw: requires: bins: - fish-audio-pp-cli install: - kind: go bins: [fish-audio-pp-cli] module: github.com/mvanhorn/printing-press-library/library/ai/fish-audio/cmd/fish-audio-pp-cli --- # Fish Audio — Printing Press CLI ## Prerequisites: Install the CLI This skill drives the `fish-audio-pp-cli` binary. **You must verify the CLI is installed before invoking any command from this skill.** If it is missing, install it first: 1. Install via the Printing Press installer. It defaults binaries to `$HOME/.local/bin` on macOS/Linux and `%LOCALAPPDATA%\Programs\PrintingPress\bin` on Windows: ```bash npx -y @mvanhorn/printing-press-library install fish-audio --cli-only ``` 2. Verify: `fish-audio-pp-cli --version` 3. Ensure the reported install directory is on `$PATH` for the agent/runtime that will invoke this skill. If the `npx` install fails (no Node, offline, etc.), fall back to a direct Go install (requires Go 1.26.6 or newer). This installs into `$GOPATH/bin` (default `$HOME/go/bin`), so add that directory to `$PATH` instead: ```bash go install github.com/mvanhorn/printing-press-library/library/ai/fish-audio/cmd/fish-audio-pp-cli@latest ``` If `--version` reports "command not found" after install, the runtime cannot see the binary directory on `$PATH`. Do not proceed with skill commands until verification succeeds. Render text to audio files, clone a voice from a sample, design a voice from a prompt, and transcribe audio, all with agent-friendly JSON output. Every render lands in a local SQLite log so you can skip duplicate renders, report spend by voice or model, and verify a clone's fidelity before it goes live. ## When to Use This CLI Use this CLI when an agent needs to turn text into audio files, clone or design a voice, transcribe audio, or account for Fish Audio spend. It is the right tool for scripted, repeatable rendering with cost tracking and for verifying a cloned voice before production use. ## Anti-triggers Do not use this CLI for: - Do not use it for live conversational voice agents or phone calls; that is Fish Audio's Agents product, out of scope here. - Do not use it for dubbing, music, or sound-effect generation; Fish Audio has no such endpoints. - Do not use it for real-time WebSocket streaming playback; use the official SDK. ## Unique Capabilities These capabilities aren't available in any other tool for this API. ### Local state that compounds - **`render log`** — See every past TTS render with its text, model, voice, byte count, and cost. _Use it to recover what was already rendered before spending credit again._ ```bash fish-audio-pp-cli render log --limit 20 --agent ``` - **`render spend`** — Total Fish Audio spend grouped by voice, model, or day for a side-by-side with your ElevenLabs invoice. _Reach for it when the question is 'what did this voice cost us this month'._ ```bash fish-audio-pp-cli render spend --group-by model --since 30d --agent ``` - **`tts render`** — Hash the request and reuse a prior identical render instead of paying for it again. _Use it in iteration loops where most lines have not changed._ ```bash fish-audio-pp-cli tts render --text "Hi, this is Pearl." --voice 802e3bc2b27e49c2995d23ef70e6ac89 --out greeting.mp3 --skip-if-rendered --agent ``` - **`render diff`** — Show the cost, model, and byte deltas between two past renders. _Use it when picking between two takes of the same line._ ```bash fish-audio-pp-cli render diff 1 2 --agent ``` ### Verify before you ship - **`voice verify`** — Render a reference phrase with a cloned voice, transcribe it back, and report word-error-rate. _Run it before swapping a production voice to a new clone._ ```bash fish-audio-pp-cli voice verify 7f92f8afb8ec43bf81429cc1c9199cb1 --agent ``` - **`tts batch`** — Estimate a batch's cost against your live API credit and refuse to start if it would overdraw. _Use it for unattended batch jobs so a long script cannot fail halfway on credit._ ```bash fish-audio-pp-cli tts batch --input lines.txt --voice 802e3bc2b27e49c2995d23ef70e6ac89 --out-dir ./out --budget-guard --agent ``` ## Command Reference **asr** — Manage asr - `fish-audio-pp-cli asr` — Speech to Text **model** — Manage model - `fish-audio-pp-cli model create` — Create Model for Users via API - `fish-audio-pp-cli model delete` — Delete Model - `fish-audio-pp-cli model get` — Get Model - `fish-audio-pp-cli model list` — List Models - `fish-audio-pp-cli model update` — Update Model **tts** — Manage tts - `fish-audio-pp-cli tts create` — Text to Speech - `fish-audio-pp-cli tts create-stream` — Text to Speech Stream with Timestamps **voice-design** — Manage voice design - `fish-audio-pp-cli voice-design` — Voice Design **wallet** — Manage wallet ### Finding the right command When you know what you want to do but not which command does it, ask the CLI directly: ```bash fish-audio-pp-cli which "" ``` `which` resolves a natural-language capability query to the best matching command from this CLI's curated feature index. Exit code `0` means at least one match; exit code `2` means no confident match — fall back to `--help` or use a narrower query. ## Recipes ### Render a greeting and check its cost ```bash fish-audio-pp-cli tts render --text "Hi, this is Pearl." --voice 802e3bc2b27e49c2995d23ef70e6ac89 --out greeting.mp3 --agent --select file,bytes,cost_usd,model ``` Writes the file and returns only the fields an agent needs to log. ### Batch a script with a budget guard ```bash fish-audio-pp-cli tts batch --input lines.txt --voice 802e3bc2b27e49c2995d23ef70e6ac89 --out-dir ./out --budget-guard --agent ``` Refuses to start if the estimated cost exceeds live API credit. ### Verify a clone before going live ```bash fish-audio-pp-cli voice verify 7f92f8afb8ec43bf81429cc1c9199cb1 --agent ``` Reports word-error-rate from a TTS-then-ASR round trip. ### Find a public voice ```bash fish-audio-pp-cli voice discover --query "calm British male" --limit 5 --agent --select id,title,languages ``` Searches the cached public catalog with FTS. ## Auth Setup Set FISH_AUDIO_API_KEY (from https://fish.audio/app/api-keys) or store it with `fish-audio-pp-cli auth set-token`. The `model` header selects the TTS model (s1, s2-pro, s2.1-pro, s2.1-pro-free); the CLI validates it because the API silently falls back to s2.1-pro on typos. Run `fish-audio-pp-cli doctor` to verify setup. ## Agent Mode Add `--agent` to any command. Expands to: `--json --compact --no-input --no-color`. - **Pipeable** — JSON on stdout, errors on stderr - **Filterable** — `--select` keeps a subset of fields. Dotted paths descend into nested structures; arrays traverse element-wise. Critical for keeping context small on verbose APIs: ```bash fish-audio-pp-cli model list --agent --select author,cover_image,created_at ``` - **Previewable** — `--dry-run` shows the request without sending - **Offline-friendly** — sync/search commands can use the local SQLite store when available - **Non-interactive** — never prompts, every input is a flag - **Explicit confirmation** — `--agent` does not imply `--yes`; pass `--yes` separately only after the target, arguments, and side effects are clear - **Explicit retries** — use `--idempotent` only when an already-existing create should count as success, and use `--ignore-missing` only when a missing delete target should count as success ### Response envelope Commands that read from the local store or the API wrap output in a provenance envelope: ```json { "meta": {"source": "live" | "local", "synced_at": "...", "reason": "..."}, "results": } ``` Parse `.results` for data and `.meta.source` to know whether it's live or local. A human-readable `N results (live)` summary is printed to stderr only when stdout is a terminal AND no machine-format flag (`--json`, `--csv`, `--compact`, `--quiet`, `--plain`, `--select`) is set — piped/agent consumers and explicit-format runs get pure JSON on stdout. ## Paths and state Agents should treat the CLI's path resolver as part of the runtime contract: - Use `--home ` for one invocation, or set `FISH_AUDIO_HOME=` to relocate all four path kinds under one root. - Use per-kind env vars only when a specific kind must diverge: `FISH_AUDIO_CONFIG_DIR`, `FISH_AUDIO_DATA_DIR`, `FISH_AUDIO_STATE_DIR`, `FISH_AUDIO_CACHE_DIR`. - Resolution order is per-kind env var, `--home`, `FISH_AUDIO_HOME`, XDG (`XDG_CONFIG_HOME`, `XDG_DATA_HOME`, `XDG_STATE_HOME`, `XDG_CACHE_HOME`), then platform defaults. - `config` contains settings like `config.toml` and profiles. `data` contains `credentials.toml`, `data.db`, cookies, and auth sidecars. `state` contains persisted queries, jobs, and `teach.log`. `cache` contains regenerable HTTP/cache files. - Stored secrets live in `credentials.toml` under the data dir. Existing legacy `config.toml` secrets are read for compatibility and leave `config.toml` on the first auth write. - Run `fish-audio-pp-cli doctor --fail-on warn` to surface path and credential-location warnings. `agent-context` exposes a schema v4 `paths` block for agents that need the resolved dirs. - For MCP, pass relocation through the MCP host config. The MCP binary does not inherit CLI flags: ```json { "mcpServers": { "fish-audio": { "command": "fish-audio-pp-mcp", "env": { "FISH_AUDIO_HOME": "/srv/fish-audio" } } } } ``` Fleet precedence: an inherited per-kind env var overrides an explicit `--home` for that kind. Use `FISH_AUDIO_HOME` or per-kind vars as durable fleet levers, and use `--home` only for a single invocation. Relocation is not reversible by unsetting env vars; move files manually before clearing `FISH_AUDIO_HOME`, or `doctor` will not find credentials left under the former root. ## Automatic learning This CLI ships a self-capturing learning loop. The CLI does its own bookkeeping: every invocation is journaled locally, a failed flag followed by a corrected retry auto-derives a `flag_alias` candidate, and a `teach` on a query family without a playbook auto-synthesizes a `playbook_candidate` from the session's journal. Your job is judgment only: `recall` first, act on surfaced candidates, `teach` the final answer, `playbook amend` when you observe a correction. You never record failures by hand. ### Step 1: `recall` before any discovery Before list/search/drill commands on a new user question, run: ```bash fish-audio-pp-cli recall "" --agent ``` The response envelope: ```json { "query": "...", "normalized": "", "query_entities": ["..."], "found": true | false, "match_score": 0.0, "results": [ { "resource_id": "...", "resource_type": "...", "venue": "...", "confidence": 2, "entity_match": "exact|partial|unknown", "source": "taught|preseed|pattern", "warnings": ["..."] } ], "mismatches": [ /* only when --debug-mismatches */ ], "warnings": [ /* top-level */ ], "candidates": [ { "id": 12, "class": "flag_alias | playbook_candidate", "summary": "...", "sightings": 3, "last_seen": "...", "rationale": "...", "next_action": ["", "fish-audio-pp-cli learnings confirm 12"] } ], "playbook": { "query_family": "...", "playbook": { "steps": [ { "cmd": "", "purpose": "..." } ], "entity_slots": ["$ENTITY"], "expected_tool_calls": 3 }, "slots_resolved": { "$ENTITY": { "token": "", "canonical": "" } }, "notes": "" }, "notes": "" } ``` Empty-store short-circuit: if the store has no learnings, playbooks, or candidates yet (recall finds nothing and `learnings list` and `learnings candidates` are both empty), skip recall for the rest of this session instead of taxing every query; resume recall-first once something has been taught. ### Step 2: decision tree Read `candidates`, `playbook`, `notes`, `results[0]`, and warnings in that order: ``` if Candidates present (warnings include "candidates_present"): -> candidates are try-then-confirm, never facts. Follow each candidate's two-step next_action verbatim: run the trial command first, then run `learnings confirm ` only after the trial verified the behavior. Reject a wrong candidate with `learnings reject `. -> NEVER re-teach something recall surfaced as a candidate; confirm or reject that candidate instead of teaching a duplicate. -> candidates ride alongside playbooks and resource hits, not instead of them; continue with the branches below after acting on them. if Playbook present: -> READ Playbook.notes verbatim FIRST (workarounds + gotchas the CLI surface doesn't expose) -> replay Playbook.steps in order, substituting Playbook.slots_resolved entries for the entity slot tokens. If a step's slot is unresolved, fall back to discovery for that step only. -> the Playbook's expected_tool_calls is a budget; if you find yourself running materially more, record the divergence via `fish-audio-pp-cli playbook amend` at end-of-session. elif Notes present (no Playbook): -> read Notes verbatim before any discovery step; they carry known gotchas for this query family even when no structured choreography exists yet. elif Found AND Results[0].EntityMatch == "exact" AND Results[0].Confidence >= 2: -> skip discovery; fetch live data for Results[*].ResourceID in parallel elif Found AND Results[0].EntityMatch == "partial": -> candidate hint, NOT a hit; read the resource title to validate before trusting elif (any row in Mismatches[] when --debug-mismatches was passed): -> treat as cold start; the stored learning is for a different entity (different canonical resolved from query_entities) else: // Found == false, no playbook, no notes -> cold start; run discovery normally; teach the answer afterward (Step 4). If the family has no playbook yet, that teach auto-synthesizes a playbook candidate from this session's journal - you do not need to record one by hand. ``` Playbook and Notes are orthogonal to the per-resource path. A recall response can carry both a Playbook AND a `Results[]` hit - use both: the Playbook tells you which choreography to run; the resource hits short-circuit specific steps. Default to skipping `mismatches`; pass `--debug-mismatches` only when investigating cold-start surprises. Candidate judgment details: `learnings confirm ` prints the candidate's full payload before materializing it - check that the printed payload matches the behavior you verified. `learnings reject ` tombstones the derivation signature so the same candidate does not resurface. The envelope carries only the few candidates worth acting on now; `fish-audio-pp-cli learnings candidates` lists the full open set. Graceful degradation: if `learnings confirm` is an unknown command, you are driving an older binary - ignore the candidates guidance and follow the rest of the protocol. ### Step 3: always read `warnings` - `low_confidence`: row exists at `confidence<2`. Treat as a hint, not a skip-discovery hit. - `resource_not_in_store`: the local store doesn't have the resource the learning points at. The match validator couldn't classify entities — direct-fetch and re-evaluate. - `cross_alias_match` (per-result): the row was taught under a different alias and matched the live query's canonical via `entity_lookups` (e.g., a "USA" teach satisfying a "United States" recall). Trust the resource_id. - `similar_shape_different_entity:` (top-level): a structurally matching row exists but its canonical entity differs from the live query's. Treated as cold start; the warning carries the conflicting canonical as a hint, but the row is NOT promoted into Results. - `ambiguous_alias` (top-level): a single query entity resolved to multiple canonicals (e.g., "Cards" → Arizona Cardinals + St. Louis Cardinals). Surface the ambiguity from context before committing to a resource. - `candidates_present` (top-level): the envelope carries a `candidates` section. Handle it via the candidates branch in Step 2 before anything else. - `lookup_refresh_available` (top-level): an entity in the query has no lookup row yet, but synced data could provide one. Run `fish-audio-pp-cli sync` to refresh entity lookups. - Top-level `no_learnings_for_query_family`: the table had no rows above the Jaccard floor. Pure cold start. ### Step 4: `teach &` after finalizing your response - always Teaching is unconditional. After resolving a query the store could not answer, background-teach the final resource mapping - no call-count threshold, no judging whether it was "worth" learning. The teach is the anchor of the loop: it triggers playbook synthesis for a family without a playbook, and same-referent phrasings fold into one family so near-duplicate teaches do not fragment the store. Fire it after assembling your user-facing response but BEFORE emitting it, with a shell `&` so the call returns immediately: ```bash fish-audio-pp-cli teach --query "" --resource-type --resource --resource # (append shell `&` to background it) ``` Silent on success. Errors only land in `teach.log` under the resolved state dir. Teach the **most specific** resource - if the user asked a broad question and you walked through parent records to find the specific answer, teach the leaf id, not the parent. The CLI uses seeded `entity_lookups` for cross-alias resolution at recall time, so a teach under one alias (e.g., "Niners") satisfies future queries under another alias (e.g., "49ers", "San Francisco") automatically. PII rule: teach the structural question with identifiers stripped - never include names, emails, phone numbers, account ids, or other personal identifiers in taught queries or notes. The CLI scans teach queries for obvious email/phone shapes and warns, but does not block; strip before teaching rather than relying on the warning. ### Step 5: playbooks - optional flags, automatic synthesis You do not need to decide whether a session "deserves" a playbook: a teach on a family without one auto-synthesizes a `playbook_candidate` from the session's journal, and the next session judges it via confirm/reject. Attach explicit playbook flags only when you already hold choreography worth recording verbatim - workarounds the CLI didn't surface (silently-dropped flags, undocumented params, pagination tricks, payload gotchas). Prefer the **integrated one-call form** - record the resource learning and the playbook in the same `teach` invocation: ```bash # Common case: record both the resource learning AND the playbook in one call. fish-audio-pp-cli teach \ --query "" \ --re