feat(source): add source@v1 capture-and-stream client

Add a 'sendspin source' command that runs a source client: capture audio
from a local input (line-in/microphone) or a sine test tone, encode it
(PCM, FLAC, or Opus via PyAV), and stream it to a server, which mixes and
distributes it to players. The server controls when the source streams.

- source_utils: PyAV SourceEncoder (exposes raw codec header) + RMS level
- source_stream: SourceStreamer capture loop (sounddevice / sine), server
  start/stop handling, optional line-sense signal reporting
- cli: 'source' subcommand and 'audio-devices inputs' listing
- audio_devices: input-device enumeration
- settings: SourceSettings persistence (settings-source.json)
- README: source mode documentation

Author: rudyberends

README.md

@@ -296,6 +296,46 @@ Use `--manufacturer` and `--product-name` to override the device identity report
 sendspin daemon --name "Living Room" --manufacturer "Acme" --product-name "Living Room Speaker"
 ```
 
+### Source Mode
+
+Run as a **source** client to capture audio from a local input (line-in, turntable,
+Bluetooth receiver, or microphone) and stream it *into* Sendspin. The server mixes
+and distributes it to players like any other audio, so an analog input on one machine
+can play back synchronized across the whole group.
+
+```bash
+# Capture the default input device and stream it to a discovered server
+sendspin source
+
+# Pick a specific server, input device, and codec
+sendspin source --url ws://192.168.1.50:8927/sendspin --device 2 --codec flac
+
+# Stream a 440 Hz sine test tone (no capture hardware needed)
+sendspin source --input sine
+```
+
+List available capture devices:
+
+```bash
+sendspin audio-devices inputs
+```
+
+Key options:
+
+- `--input {linein,sine}` — capture from a real input device, or generate a sine test tone. Defaults to `linein`; passing `--device` implies `linein`.
+- `--device` — input device index or name (see `audio-devices inputs`).
+- `--codec {pcm,opus,flac}` — codec used to encode captured audio before sending (default `pcm`). Capture is 16-bit.
+- `--sample-rate` / `--channels` — capture format (default 48000 Hz, 2 channels).
+- `--line-sense` — report input signal presence to the server via `client/state`; the server may use it to decide when to start/stop the source.
+
+The **server** decides when a source streams: a source stays idle until the server
+sends a `start` command, and stops on `stop` or disconnect. A device may run both the
+`source` and `player` roles; when it does, it never plays its captured input locally —
+it only plays back what the server distributes, staying in sync with the group.
+
+Source-mode preferences (client id, last server, input/codec defaults) are persisted
+to `~/.config/sendspin/settings-source.json`.
+
 ### Hooks
 
 You can run external commands when audio streams start or stop. This is useful for controlling amplifiers, lighting, or other home automation:

sendspin/audio_devices.py

@@ -80,6 +80,38 @@ def query_devices() -> list[AudioDevice]:
     return result
 
 
+@dataclass(slots=True)
+class InputDevice:
+    """Represents an audio input (capture) device."""
+
+    index: int
+    name: str
+    input_channels: int
+    sample_rate: float
+    is_default: bool
+
+
+def query_input_devices() -> list[InputDevice]:
+    """Query all available audio input (capture) devices."""
+    devices = sounddevice.query_devices()
+    default_input = int(sounddevice.default.device[0])
+
+    result: list[InputDevice] = []
+    for i in range(len(devices)):
+        dev = devices[i]
+        if dev["max_input_channels"] > 0:
+            result.append(
+                InputDevice(
+                    index=i,
+                    name=str(dev["name"]),
+                    input_channels=int(dev["max_input_channels"]),
+                    sample_rate=float(dev["default_samplerate"]),
+                    is_default=(i == default_input),
+                )
+            )
+    return result
+
+
 def _check_format(device: AudioDevice, rate: int, channels: int, dtype: str) -> bool:
     """Check if a specific audio format is supported by the device."""
     try: