Skip to content

Quick Start

The real HTTP recognition backend uses CHATVOICE_ASR_API_URL; see configuration for endpoint and credential setup.

Choose a goal first. A model-free trial checks the interface, not transcription or provider readiness.

  • Use an existing deployment

    Open the Speakr URL supplied by its operator, then choose account or guest mode.

    Web guide

  • Try the interface

    Use stub-local without GPU models or cloud credentials.

    UI trial

  • Run a real service

    Configure real ASR and text models, then enable optional voice capabilities.

    Real service

Install

Python 3.10+ is required. This POSIX example uses a virtual environment under ChatArch home. Windows uses the same package with its platform-specific activation command.

python3 -m venv "$HOME/.chatarch/chatvoice/.venv"
. "$HOME/.chatarch/chatvoice/.venv/bin/activate"
python -m pip install "ChatVoice[web]==0.4.1"
chatvoice --version
chatvoice --tree

The web extra does not install CUDA, PyTorch or FunASR models.

UI trial

export CHATVOICE_ASR_CHANNEL=stub-local
chatvoice serve app --dry-run --json
chatvoice serve app --host 127.0.0.1 --port 18087

Open http://127.0.0.1:18087/ on the service machine and choose guest mode. Stop the foreground server with the terminal's normal interrupt.

Not real transcription

The stub channel loads no model. Summaries, Todo, TTS and realtime still require their own configured backends. A remote user's loopback address is not the server: use the deployment's HTTPS entry.

Real service

Capability Required preparation
ASR HTTP endpoint, or working local FunASR model and device in the service environment
Notes, revisions, Todo Notes base URL, key and model
Automatic titles Title base URL, key and model
System TTS Protocol, endpoint, key, model, voices; resource ID for applicable protocols
Voice cloning Reachable VoiceClone service; account and voice consent for generation
Realtime Current Qwen realtime backend credentials and model entitlement

Follow configuration, then run:

chatenv test -t chatvoice -I
chatvoice serve app --host 127.0.0.1 --port 18087

The ChatEnv probe calls configured text/TTS backends and may consume quota. It does not fully validate ASR, realtime or voice cloning.

Invite an account

Public registration is disabled. Set the password securely in CHATVOICE_ACCOUNT_LOGIN within the service environment, then run:

chatvoice accounts add member@example.com --display-name Member --password-env CHATVOICE_ACCOUNT_LOGIN --json
chatvoice accounts list --json

Never place the password in command arguments or public scripts. Sign in through the web app afterward. See deployment for persistent supervision and HTTPS.