spych 4.0.0

Communicate with your favorite AI model by talking to it.

Spych

Spych (pronounced "speech"): Talk with your computer like it's your personal assistant without sending your voice to the cloud.

A lightweight, fully offline Python toolkit for wake word detection, audio transcription, spoken AI responses, and AI integrations. Built on faster-whisper, PvRecorder, and Kokoro.

API Docs: https://connor-makowski.github.io/spych/spych.html

---

Installation

Recommended: pipx

pipx install spych

Alternative: pip

pip install spych

TTS Extras

By default, Spych automatically installs the right TTS backend for your Python version. You can also install explicitly:

pipx install "spych[kokoro]"       # Fast, lightweight (Python < 3.13 recommended)
pipx install "spych[chatterbox]"   # High-quality voice cloning (Python >= 3.13 required)

---

Quick Start

# Navigate to your project directory first
cd ~/my_project

# Voice-control Claude Code — say "hey claude" to trigger
spych claude

# Use a personality preset — say "hey jarvis" to trigger
spych claude --personality jarvis

# Voice-control a local Ollama model — say "hey llama" to trigger
spych ollama --model llama3.2:latest

💡 Pro tip: Saying "Hey Claude" or "Hey Llama" tends to trigger more reliably than the bare wake word.

Say "terminate" (or press Ctrl+C) to stop any session.

---

CLI

Available Agents

All agents require their respective CLI tool to be installed and authenticated before use.

Command	Alias	Description	Default wake words
spych claude_code_cli	—	Voice-control Claude Code via the CLI	claude, clod, cloud, clawed
spych claude_code_sdk	spych claude	Voice-control Claude Code via the Agent SDK	claude, clod, cloud, clawed
spych codex_cli	spych codex	Voice-control the OpenAI Codex agent	codex
spych gemini_cli	spych gemini	Voice-control the Google Gemini agent	gemini, google
spych opencode_cli	spych opencode	Voice-control the OpenCode agent	opencode, open code
spych ollama	—	Talk to a local Ollama model	llama, ollama, lama

Available Utilities

The following utilities are also available as CLI commands. They don't use wake words, but serve various auxiliary functions like live transcription and voice profiling.

Command	Description
spych --version	Print the version number and exit
spych --help	Show detailed usage instructions and exit
spych live	Continuous speech-to-text transcription to file
spych multi	Run multiple agents simultaneously
spych profile_my_voice	Record a voice sample for TTS cloning

Global Flags

These must be placed before the agent name:

spych --theme light claude

Flag	Options	Default	Description
--theme	dark, light, solarized, mono	dark	Terminal colour theme

---

Common Flags

All agent subcommands accept these flags:

Flag	Default	Description
--personality NAME	—	Apply a named preset (sets wake words, voice, name, style)
--name NAME	(agent default)	Custom display name shown in the terminal
--wake-words WORD [...]	(agent default)	One or more words that trigger the agent
--terminate-words WORD [...]	terminate	Words that stop the listener
--listen-duration SECONDS	0 (VAD auto)	Seconds to record after wake word
--follow-up-listen-duration SECONDS	0	Seconds to listen for a follow-up answer
--inactivity-timeout SECONDS	4.0	Seconds of silence before returning to wake word
--use-speaker BOOL	true	Speak responses aloud via TTS
--speaker-voice VOICE	af_heart	Voice name for spoken responses
--speaker-backend BACKEND	(auto)	chatterbox or kokoro
--response-style STYLE	—	Style preset or custom instruction for spoken output

Coding agents (claude, codex, gemini, opencode) also accept:

Flag	Default	Description
--continue-conversation BOOL	true	Resume the most recent session
--show-tool-events BOOL	true	Print live tool start/end events

Agent-specific flags:

Agent	Flag	Default	Description
ollama	--model	llama3.2:latest	Ollama model name
ollama	--history-length	10	Past interactions to include in context
ollama	--host	http://localhost:11434	Ollama instance URL
opencode_cli	--model	—	Model in provider/model format
claude_code_sdk	--setting-sources	user project local	Claude Code settings sources

---

Personalities

Personalities are named presets that bundle a wake word list, voice, display name, and response style into a single flag. Any explicit flag overrides the preset.

spych claude --personality jarvis
# equivalent to:
spych claude --name "J.A.R.V.I.S." --wake-words jarvis jarves \
             --speaker-voice bm_george --use-speaker true \
             --response-style jarvis

Name	Wake words	Voice	Style
assistant	assistant, helper, computer	af_heart	assistant — helpful, precise, informative
friend	friend, buddy, pal	af_amy	friendly — warm and simple
jarvis	jarvis, jarves, jargus, jervis	bm_george	jarvis — precise, dry wit, "sir"
pirate	blackbeard, pirate, ahoy	am_michael	pirate — pirate speak, colorful
news_anchor	bella, news anchor, anchor	af_bella	news_anchor — professional broadcast tone
robot	rob, robot	am_adam	robot — monotone, literal
caveman	er, ur, caveman, cave man	am_onyx	caveman — very simple, direct

---

Response Styles

The --response-style flag shapes how the agent formats its spoken output.

Style	Description
assistant	Helpful and precise, concise and informative
concise	Key points only, direct
friendly	Warm, approachable, simple language
military	Brevity-style, short sentences
five_year_old	Simple words, very short
fast	As brief as reasonably possible
pirate	Pirate speak, colorful
news_anchor	Professional broadcast tone
haiku	5-7-5 haiku form
shakespearean	Elizabethan English
robot	Monotone, literal
caveman	Very simple, direct
yoda	Inverted sentence structure
jarvis	J.A.R.V.I.S. from Iron Man — precise, dry wit, addresses user as "sir"

You can also pass any custom instruction string directly: --response-style "Reply in exactly one sentence.".

---

Text-to-Speech & Voices

Spoken responses are enabled by default for personality presets and when --use-speaker true is set.

spych claude --use-speaker true --speaker-voice bm_george
spych claude --use-speaker true --speaker-backend kokoro
spych claude --use-speaker false   # disable TTS

When TTS is active, short responses are spoken verbatim; longer ones use the agent's short summary. If the response ends with a question, Spych automatically listens for a follow-up — no wake word required.

TTS Backends

Backend	Best for	Python support
Chatterbox (default priority)	Natural voices, zero-shot voice cloning	3.11+ (required for 3.13+)
Kokoro (lightweight fallback)	Fast, low-resource devices (e.g. Raspberry Pi)	3.11–3.12 recommended

Spych tries Chatterbox first, then Kokoro. Use --speaker-backend to force one explicitly.

Available Voices

The same voice names work for both backends.

• Chatterbox wave voices: https://github.com/connor-makowski/spych/tree/main/voices/wave
• Kokoro pt voices (56 total): https://github.com/connor-makowski/spych/tree/main/voices/pt

American English (am_ / af_):

Voice	Gender	Grade
af_heart	F	A (default)
af_bella	F	A-
af_nicole	F	B-
am_michael	M	C+
am_fenrir	M	C+
am_puck	M	C+

British English (bm_ / bf_):

Voice	Gender	Grade
bf_emma	F	B-
bf_isabella	F	C
bm_george	M	C

Voice Cloning

Record a 10-second sample of your voice, then use it as the speaker voice. Requires the Chatterbox backend.

# Step 1: record your profile
spych profile_my_voice --name my_voice

# Step 2: use it
spych claude --use-speaker true --speaker-voice my_voice --speaker-backend chatterbox

# Or use any .wav file directly
spych claude --use-speaker true --speaker-voice /path/to/my_voice.wav --speaker-backend chatterbox

---

Live Transcription

spych live continuously records from the microphone using VAD and writes the transcript to disk in real time. No wake word required — it transcribes everything until stopped.

CLI

spych live                                                 # writes transcript.srt
spych live --output-path meeting --output-format both
spych live --terminate-words "stop recording"
spych live --no-timestamps --whisper-model small.en

Stop by pressing the stop key (default: q + Enter), saying a terminate word, or pressing Ctrl+C.

Parameters

Flag	Default	Description
--output-path PATH	transcript	Base output file path without extension
--output-format FORMAT	srt	txt, srt, or both
--no-timestamps	false	Omit timestamps from terminal and .txt output
--stop-key KEY	q	Key (then Enter) to stop the session
--terminate-words WORD [...]	—	Spoken words that stop the session
--device-index N	-1	Microphone device index; -1 uses system default
--whisper-model MODEL	base.en	faster-whisper model name
--whisper-device DEVICE	cpu	cpu or cuda
--whisper-compute-type TYPE	int8	int8, float16, or float32
--no-speech-threshold FLOAT	0.3	Whisper segments above this no_speech_prob are dropped
--speech-threshold FLOAT	0.5	VAD speech onset probability
--silence-threshold FLOAT	0.35	VAD silence probability during speech
--silence-frames N	20	Consecutive silent frames to end a segment (~32ms each)
--speech-pad-frames N	5	Pre-roll frames and onset confirmation count
--max-speech-duration SECONDS	30.0	Hard cap on a single segment
--context-words N	32	Trailing words passed as whisper initial_prompt

Python

from spych.live import SpychLive

SpychLive(
    output_format="srt",          # "txt", "srt", or "both"
    output_path="my_transcript",  # written to my_transcript.srt
    show_timestamps=True,
    stop_key="q",                 # type q + Enter to stop
    terminate_words=["stop recording"],
).start()

SpychLive Parameters

Parameter	Default	Description
output_format	"srt"	Output format(s): "txt", "srt", or "both"
output_path	"transcript"	Base path without extension
show_timestamps	True	Prepend [HH:MM:SS] timestamps to terminal and .txt output
stop_key	"q"	Key (then Enter) to stop the session
terminate_words	None	Spoken words that stop the session
on_terminate	None	No-argument callback executed when a terminate word fires
device_index	-1	Microphone device index; -1 uses system default
whisper_model	"base.en"	faster-whisper model name
whisper_device	"cpu"	Device for inference: "cpu" or "cuda"
whisper_compute_type	"int8"	Compute precision: "int8", "float16", or "float32"
no_speech_threshold	0.4	Whisper segments above this are discarded
speech_threshold	0.5	Silero VAD onset probability
silence_threshold	0.35	Silero VAD silence probability during speech
silence_frames_threshold	20	Consecutive silent frames to close a segment
speech_pad_frames	5	Pre-roll frame count and onset confirmation threshold
max_speech_duration_s	30.0	Hard cap on a single segment in seconds
context_words	32	Trailing transcript words passed as initial_prompt

---

Multi-agent

Run several agents simultaneously under a single listener, each bound to its own wake words. Say "hey claude" to talk to Claude, "hey llama" to talk to Ollama — all in the same terminal session.

CLI

# Two agents, default wake words
spych multi --agents claude gemini

# Include Ollama with a specific model
spych multi --agents claude ollama --ollama-model llama3.2:latest

# Tune listen duration across all agents
spych multi --agents claude codex --listen-duration 8

Multi-agent CLI Flags

Flag	Default	Description
--agents AGENT [...]	(required)	Agents to run: claude (claude_code_cli), claude_sdk (claude_code_sdk), codex (codex_cli), gemini (gemini_cli), opencode (opencode_cli), ollama
--terminate-words WORD [...]	terminate	Words that stop all agents
--listen-duration SECONDS	5	Seconds to listen after a wake word
--follow-up-listen-duration SECONDS	0	Seconds to listen for follow-up answers
--inactivity-timeout SECONDS	4.0	Seconds of silence before returning to wake word
--continue-conversation BOOL	true	Resume the most recent session for each coding agent
--show-tool-events BOOL	true	Print live tool start/end events
--use-speaker BOOL	true	Speak responses aloud via TTS
--speaker-backend BACKEND	(auto)	chatterbox or kokoro
--ollama-model MODEL	llama3.2:latest	Only used when ollama is in --agents
--ollama-host URL	http://localhost:11434	Only used when ollama is in --agents
--ollama-history-length N	10	Only used when ollama is in --agents
--opencode-model MODEL	—	provider/model format. Only used when opencode_cli is in --agents
--setting-sources SOURCE [...]	user project local	Only used when claude_code_sdk is in --agents

Python

from spych.core import Spych
from spych.orchestrator import SpychOrchestrator
from spych.agents.claude import LocalClaudeCodeCLIResponder
from spych.agents.ollama import OllamaResponder

spych_object = Spych(whisper_model="base.en")

SpychOrchestrator(
    entries=[
        {
            "responder": LocalClaudeCodeCLIResponder(spych_object=spych_object),
            "wake_words": ["claude", "clod", "cloud", "clawed"],
            "terminate_words": ["terminate"],
        },
        {
            "responder": OllamaResponder(spych_object=spych_object, model="llama3.2:latest"),
            "wake_words": ["llama", "ollama", "lama"],
        },
    ]
).start()

OrchestratorEntry Keys

Key	Required	Default	Description
responder	✓	—	A BaseResponder instance
wake_words	✓	—	Words that trigger this responder. Must be unique across all entries
terminate_words		["terminate"]	Words that stop the entire orchestrator

SpychOrchestrator Parameters

Parameter	Default	Description
entries	(required)	List of OrchestratorEntry dicts
spych_wake_kwargs	None	Extra kwargs forwarded to SpychWake

---

Python — Built-in Agents

The same agents available from the CLI can be used directly from Python.

Claude Code CLI

from spych.agents import claude_code_cli

# Say "hey claude" to trigger
claude_code_cli()

Claude Code SDK

from spych.agents import claude_code_sdk

# Say "hey claude" to trigger
claude_code_sdk()

Codex CLI

from spych.agents import codex_cli

# Say "hey codex" to trigger
codex_cli()

Gemini CLI

from spych.agents import gemini_cli

# Say "hey gemini" to trigger
gemini_cli()

OpenCode CLI

from spych.agents import opencode_cli

# Say "hey opencode" to trigger
opencode_cli()

Ollama

from spych.agents import ollama

# Pull the model first: ollama pull llama3.2:latest
# Say "hey llama" to trigger
ollama(model="llama3.2:latest")

Coding Agent Parameters

Parameter	claude_code_cli	claude_code_sdk	codex_cli	gemini_cli	opencode_cli	Description
name	Claude	Claude	Codex	Gemini	OpenCode	Custom display name
wake_words	["claude", "clod", "cloud", "clawed"]	["claude", "clod", "cloud", "clawed"]	["codex"]	["gemini", "google"]	["opencode", "open code"]	Words that trigger the agent
terminate_words	["terminate"]	["terminate"]	["terminate"]	["terminate"]	["terminate"]	Words that stop the listener
model	—	—	—	—	None	Model in provider/model format
listen_duration	0	0	0	0	0	Seconds to listen (0 = VAD auto)
continue_conversation	True	True	True	True	True	Resume the most recent session
setting_sources	—	["user", "project", "local"]	—	—	—	Claude Code settings sources
show_tool_events	True	True	True	True	True	Print live tool start/end events
use_speaker	False	False	False	False	False	Speak responses aloud via TTS
speaker_voice	"af_heart"	"af_heart"	"af_heart"	"af_heart"	"af_heart"	Voice name for TTS
response_style	""	""	""	""	""	Style preset or custom instruction
spych_kwargs	—	—	—	—	—	Extra kwargs passed to Spych
spych_wake_kwargs	—	—	—	—	—	Extra kwargs passed to SpychWake

Ollama Parameters

Parameter	Default	Description
name	"Ollama"	Custom display name
wake_words	["llama", "ollama", "lama"]	Words that trigger the agent
terminate_words	["terminate"]	Words that stop the listener
model	"llama3.2:latest"	Ollama model name
listen_duration	0	Seconds to listen (0 = VAD auto)
history_length	10	Past interactions to include in context
host	"http://localhost:11434"	Ollama instance URL
use_speaker	False	Speak responses aloud via TTS
speaker_voice	"af_heart"	Voice name for TTS
response_style	""	Style preset or custom instruction
spych_kwargs	None	Extra kwargs passed to Spych
spych_wake_kwargs	None	Extra kwargs passed to SpychWake

---

Python: Building Your Own Agent

Subclass BaseResponder, implement respond, and Spych handles the rest: wake word detection, transcription, spinner UI, timing, TTS, error handling.

respond() must return an AgentResponse. Use self.format_prompt() to inject the JSON schema into your prompt and self.parse_output() to parse the result:

from spych.responders import BaseResponder, AgentResponse

class MyResponder(BaseResponder):
    def respond(self, user_input: str) -> AgentResponse:
        raw = call_my_llm(self.format_prompt(user_input))
        return self.parse_output(raw)

A complete working example with a custom wake word:

from spych import Spych, SpychOrchestrator
from spych.responders import BaseResponder, AgentResponse

class EchoResponder(BaseResponder):
    def respond(self, user_input: str) -> AgentResponse:
        return AgentResponse(
            response=f"'{self.name}' heard: {user_input}",
            summary=f"Heard: {user_input}",
            requires_user_feedback=False,
        )

SpychOrchestrator(
    entries=[
        {
            "responder": EchoResponder(
                spych_object=Spych(whisper_model="base.en"),
                listen_duration=5,
                name="TestResponder",
            ),
            "wake_words": ["test"],
            "terminate_words": ["terminate"],
        }
    ]
).start()

You can also subclass a built-in agent. For example, a translation agent that routes to Ollama:

from spych import Spych, SpychOrchestrator
from spych.agents import OllamaResponder
from spych.responders import AgentResponse

class Spanish(OllamaResponder):
    def respond(self, user_input: str) -> AgentResponse:
        user_input = f"Translate the following to Spanish and return only the translated text: '{user_input}'"
        return super().respond(user_input)

class German(OllamaResponder):
    def respond(self, user_input: str) -> AgentResponse:
        user_input = f"Translate the following to German and return only the translated text: '{user_input}'"
        return super().respond(user_input)

spych_object = Spych(whisper_model="base.en")

SpychOrchestrator(
    entries=[
        {
            "responder": Spanish(spych_object=spych_object, name="SpanishTranslator", model="llama3.2:latest"),
            "wake_words": ["spanish"],
            "terminate_words": ["terminate"],
        },
        {
            "responder": German(spych_object=spych_object, name="GermanTranslator", model="llama3.2:latest"),
            "wake_words": ["german"],
            "terminate_words": ["terminate"],
        },
    ]
).start()

Think your agent would be useful to others? Open a PR or file a feature request via a GitHub issue.

---

Python: Lower-Level API

Need more control? Use Spych and SpychWake directly.

Transcription

from spych import Spych

spych = Spych(
    whisper_model="base.en",  # tiny, small, medium, large — all faster-whisper models work
    whisper_device="cpu",     # use "cuda" for Nvidia GPU
)

print(spych.listen(duration=5))

See: https://connor-makowski.github.io/spych/spych/core.html

Wake Word Detection

from spych import SpychWake, Spych

spych = Spych(whisper_model="base.en", whisper_device="cpu")

def on_wake():
    print("Wake word detected! Listening...")
    print(spych.listen(duration=5))

SpychWake(
    wake_word_map={"speech": on_wake},
    whisper_model="tiny.en",
    whisper_device="cpu",
).start()

See: https://connor-makowski.github.io/spych/spych/wake.html

---

API Reference

Full docs including all parameters and methods: https://connor-makowski.github.io/spych/spych.html

---

Support

Found a bug or want a new feature? Open an issue on GitHub.

---

Contributing

Contributions are welcome!

1. Fork the repo and clone it locally.
2. Make your changes.
3. Run tests and make sure they pass.
4. Commit atomically with clear messages.
5. Submit a pull request.

Virtual environment setup:

python3.11 -m venv venv
source venv/bin/activate
pip install -r requirements.txt
./utils/test.sh