Keep streaming TTS open; name the swappable-provider pattern

Utterances cross as chunks with an explicit terminator, so whole-utterance is
a stream of one chunk and a streaming TTS needs no protocol change. Records
what must not be built — a single-body download-then-play player — since that
forecloses streaming silently.

Also notes the transport is fully exercisable with espeak-ng or a canned WAV,
so audio transport need not wait on GPT-SoVITS.

The provider-behind-an-interface shape has now been picked three times
independently, so it is named as the project idiom rather than re-derived.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013kj58Jugkic783b8ZrtTVZ

docs/ARCHITECTURE.md

@@ -86,6 +86,25 @@ Topic names and payload schemas are a **contract shared across repos** — Andro
 and ESP32 clients depend on them — so they are versioned in their own repo
 rather than duplicated per client.
 
+### Swappable providers
+
+The same shape has now been chosen independently three times, so it is the
+project's idiom rather than three coincidences:
+
+| Component | Interface | Real | Trivial |
+|---|---|---|---|
+| `svc-nav` | routing provider | local Valhalla | cloud routing API |
+| `svc-gw` | input source | the bus | synthetic state |
+| voice | TTS provider | GPT-SoVITS | `espeak-ng`, or a canned WAV |
+
+The rule in each case: **one implementation of the contract, a swappable source
+behind it.** That is what lets a service mock itself without a second codebase to
+drift (decision 19), and what keeps a heavyweight dependency from blocking work on
+everything downstream of it.
+
+A new component that has an expensive or not-yet-existing dependency should assume
+this shape rather than rediscovering it.
+
 ## Data flow and direction
 
 ### Inside the Orin, direction belongs to topics, not links
@@ -279,7 +298,7 @@ rewriting the document.
 | 9 | Nav guidance published on the bus; route requests over HTTP via `telemetry-gw` | **Proposed** | 2026-09-03 |
 | 10 | Map artifacts delivered by `rsync` over LAN only; Valhalla tiles kept as a directory, not a tar | Accepted | 2026-09-03 |
 | 11 | Boot ordering is capture-first: acquisition reaches ready before serving or voice | Accepted | 2026-09-03 |
-| 12 | The **tablet** owns the car's aux input; TTS crosses as discrete utterances, not a stream | Accepted | 2026-09-04 |
+| 12 | The **tablet** owns the car's aux input; TTS crosses as chunked utterances, streaming-capable | Accepted | 2026-09-04 |
 | 13 | Reasoning stays in the cloud — a local LLM does not fit beside the voice stack | Accepted | 2026-09-03 |
 | 14 | Recordings are bounded with rotation | Accepted | 2026-09-04 |
 | 15 | Boundary is two simplex channels — lossy stream down, acknowledged request/response up | Accepted | 2026-09-03 |

docs/ORIN-RUNTIME.md

@@ -138,17 +138,35 @@ The tablet is the media source, so it holds the aux cable whatever is decided
 about speech. Giving the Orin its own aux line would not free the display unit —
 it would add a second permanent cable to the car and leave the first one in place.
 
-### TTS crosses as utterances, not as a stream
+### TTS crosses as chunked utterances
 
-An utterance is short, finite, and fully synthesised before it is needed. So it
+An utterance travels as a **sequence of chunks with an explicit terminator**, over
-ships as a **discrete object over the reliable channel**, not as a real-time audio
+the reliable channel. Not as one object, and not as a clock-synchronised audio
-stream: no jitter buffer, no underrun handling, no clock synchronisation, and a
+stream.
-dropped packet is a retry rather than an audible gap.
 
-The cost is that speech begins only once synthesis *and* transfer complete. For a
+This is deliberately the shape that keeps both modes open, because whole-utterance
-car assistant that is fine — nothing here is lip-synced. If streaming TTS is ever
+is simply a stream of one chunk:
-wanted, so that speech starts before synthesis finishes, this is the decision to
+
-revisit.
+| TTS capability | What it sends |
+|---|---|
+| Synthesises fully, then returns | one chunk, then terminator |
+| Streams as it synthesises | chunks as produced, then terminator |
+
+Same protocol, same receiver. The receiver starts playback on the first chunk,
+appends as more arrive, and finishes on the terminator — which is what Android's
+`AudioTrack` streaming mode does natively.
+
+**What must not be built:** an utterance as a single response body with a
+"download the file, then play the file" player. That works today and forecloses
+streaming permanently. Whether the TTS we land on streams is still open, so the
+transport must not assume either way.
+
+### The transport is testable with no real TTS
+
+Audio is the lowest-priority part of the system, but its transport does not have
+to wait for GPT-SoVITS. A canned WAV, or `espeak-ng` — instant, tiny, and awful —
+exercises chunking, ordering, the terminator, ducking and the no-sink path
+completely. Voice quality is irrelevant to every one of those.
 
 ### Ducking is the tablet's job