From 0be7c1024df61281ddf984798fe96e72604841d0 Mon Sep 17 00:00:00 2001
From: Jer Miller <jeremie.miller@gmail.com>
Date: Thu, 9 Jul 2026 14:19:55 -0600
Subject: [PATCH] fix(speakers): resolve segment audio by registered format,
 not hardcoded .flac

The speakers review, per-sentence play, and owner/discovery sample players were dead on any journal whose segments store audio.m4a (every mac-captured journal), because every audio URL-building site hardcoded {source}.flac.

Resolution now probes {source}.<ext> over solstone.think.media.FORMATS order (.flac first, preserving existing behavior when both exist), via one helper in the new leaf module solstone/apps/speakers/audio.py. Both duplicated _audio_url builders are deleted.

serve_audio now serves the MIME type from media.MIME_TYPES and passes conditional=True, so browser range/seek requests work.

Payload shape change: api_review gains an audio_mimetype key, so tests/baselines/api/speakers/review.json was regenerated via make update-api-baselines (not hand-edited). This is the one API-visible change.

Owner/discovery samples now emit audio_url: null when a segment's audio has been purged (embeddings outlive audio), and the three <audio> renders omit the element rather than emitting src="None".
---
 solstone/apps/speakers/audio.py               | 49 ++++++++++
 solstone/apps/speakers/discovery.py           | 10 +-
 solstone/apps/speakers/owner.py               | 10 +-
 solstone/apps/speakers/routes.py              | 17 +++-
 solstone/apps/speakers/tests/conftest.py      |  9 +-
 solstone/apps/speakers/tests/test_audio.py    | 74 ++++++++++++++
 .../apps/speakers/tests/test_discovery.py     | 47 ++++++++-
 solstone/apps/speakers/tests/test_owner.py    | 75 +++++++++++++++
 solstone/apps/speakers/tests/test_routes.py   | 96 +++++++++++++++++++
 solstone/apps/speakers/workspace.html         | 10 +-
 tests/baselines/api/speakers/review.json      |  1 +
 11 files changed, 370 insertions(+), 28 deletions(-)
 create mode 100644 solstone/apps/speakers/audio.py
 create mode 100644 solstone/apps/speakers/tests/test_audio.py

diff --git a/solstone/apps/speakers/audio.py b/solstone/apps/speakers/audio.py
new file mode 100644
index 000000000..4afaf0704
--- /dev/null
+++ b/solstone/apps/speakers/audio.py
@@ -0,0 +1,49 @@
+# SPDX-License-Identifier: AGPL-3.0-only
+# Copyright (c) 2026 sol pbc
+
+"""Audio file resolution helpers for the speakers app."""
+
+from __future__ import annotations
+
+from pathlib import Path
+
+from solstone.think.media import FORMATS
+from solstone.think.utils import segment_path
+
+__all__ = ["audio_serve_url", "resolve_audio_file", "resolve_audio_url"]
+
+_ORDERED_AUDIO_EXTENSIONS: tuple[str, ...] = tuple(
+    ext for ext, _mime, kind in FORMATS if kind == "audio"
+)
+
+
+def resolve_audio_file(segment_dir: Path, source: str) -> Path | None:
+    """Return the registered audio file for source in segment_dir, if present.
+
+    Only checks {source}.<registered-audio-extension> files directly inside
+    segment_dir, in solstone.think.media.FORMATS order.
+    """
+    for suffix in _ORDERED_AUDIO_EXTENSIONS:
+        path = segment_dir / f"{source}{suffix}"
+        if path.is_file():
+            return path
+    return None
+
+
+def audio_serve_url(day: str, stream: str, segment_key: str, filename: str) -> str:
+    """Return the speakers audio-serving URL for a segment audio filename."""
+    return f"/app/speakers/api/serve_audio/{day}/{stream}/{segment_key}/{filename}"
+
+
+def resolve_audio_url(
+    day: str,
+    stream: str,
+    segment_key: str,
+    source: str,
+) -> str | None:
+    """Resolve the speakers audio-serving URL for a segment source, if present."""
+    segment_dir = segment_path(day, segment_key, stream, create=False)
+    audio_path = resolve_audio_file(segment_dir, source)
+    if audio_path is None:
+        return None
+    return audio_serve_url(day, stream, segment_key, audio_path.name)
diff --git a/solstone/apps/speakers/discovery.py b/solstone/apps/speakers/discovery.py
index 7451f071c..e409504f7 100644
--- a/solstone/apps/speakers/discovery.py
+++ b/solstone/apps/speakers/discovery.py
@@ -12,6 +12,7 @@ from datetime import datetime
 from pathlib import Path
 from typing import Any
 
+from solstone.apps.speakers.audio import resolve_audio_url
 from solstone.think.journal_io import atomic_replace
 from solstone.think.utils import day_dirs, day_path, get_journal, now_ms, segment_path
 
@@ -49,11 +50,6 @@ def _owner_helpers():
     return load_owner_centroid
 
 
-def _audio_url(day: str, stream: str, segment_key: str, source: str) -> str:
-    """Build the existing speakers audio-serving URL for a sample."""
-    return f"/app/speakers/api/serve_audio/{day}/{stream}/{segment_key}/{source}.flac"
-
-
 def _discovery_cache_path() -> Path:
     """Return the temporary cache path for discovery cluster assignments."""
     awareness_dir = Path(get_journal()) / "awareness"
@@ -255,7 +251,7 @@ def discover_unknown_speakers() -> dict[str, Any]:
             samples.append(
                 {
                     **record,
-                    "audio_url": _audio_url(
+                    "audio_url": resolve_audio_url(
                         record["day"],
                         record["stream"],
                         record["segment_key"],
@@ -280,7 +276,7 @@ def discover_unknown_speakers() -> dict[str, Any]:
                 )
                 sample = {
                     **record,
-                    "audio_url": _audio_url(
+                    "audio_url": resolve_audio_url(
                         record["day"],
                         record["stream"],
                         record["segment_key"],
diff --git a/solstone/apps/speakers/owner.py b/solstone/apps/speakers/owner.py
index 2ce2b28e5..8686c84e6 100644
--- a/solstone/apps/speakers/owner.py
+++ b/solstone/apps/speakers/owner.py
@@ -15,6 +15,7 @@ from pathlib import Path
 from typing import TYPE_CHECKING, Any
 
 from solstone.apps.speakers._overlap import _read_segment_overlap_fraction
+from solstone.apps.speakers.audio import resolve_audio_url
 from solstone.apps.speakers.encoder_config import (
     NOISY_FLYWHEEL_OVERLAP_MAX,
     OWNER_BOOTSTRAP_MIN_INTRA_COSINE_P25,
@@ -215,11 +216,6 @@ def _iso_now() -> str:
     return datetime.now(UTC).isoformat().replace("+00:00", "Z")
 
 
-def _audio_url(day: str, stream: str, segment_key: str, source: str) -> str:
-    """Build the existing speakers audio-serving URL for a sample."""
-    return f"/app/speakers/api/serve_audio/{day}/{stream}/{segment_key}/{source}.flac"
-
-
 def _fallback_statement_durations(jsonl_path: Path) -> dict[int, float | None]:
     """Estimate statement durations from adjacent transcript start times."""
     if not jsonl_path.exists():
@@ -967,7 +963,7 @@ def _owner_candidate_samples(
         samples.append(
             {
                 **record,
-                "audio_url": _audio_url(
+                "audio_url": resolve_audio_url(
                     record["day"],
                     record["stream"],
                     record["segment_key"],
@@ -982,7 +978,7 @@ def _owner_candidate_samples(
         record = provenance[int(position)]
         sample = {
             **record,
-            "audio_url": _audio_url(
+            "audio_url": resolve_audio_url(
                 record["day"],
                 record["stream"],
                 record["segment_key"],
diff --git a/solstone/apps/speakers/routes.py b/solstone/apps/speakers/routes.py
index 7136830b7..51c704426 100644
--- a/solstone/apps/speakers/routes.py
+++ b/solstone/apps/speakers/routes.py
@@ -33,6 +33,7 @@ from solstone.apps.speakers.attribution import (
     backfill_segments,
     save_speaker_labels,
 )
+from solstone.apps.speakers.audio import audio_serve_url, resolve_audio_file
 from solstone.apps.speakers.bootstrap import (
     bootstrap_voiceprints,
     link_import,
@@ -107,6 +108,7 @@ from solstone.think.entities.journal import (
 )
 from solstone.think.journal_io.errors import LockTimeout
 from solstone.think.journal_io.npz import load_npz, update_npz
+from solstone.think.media import MIME_TYPES
 from solstone.think.utils import (
     STREAM_RE,
     day_dirs,
@@ -967,10 +969,11 @@ def api_review(day: str, stream: str, segment_key: str, source: str) -> Any:
     all_entities.sort(key=lambda x: (not x["is_principal"], x["name"].lower()))
 
     audio_file = None
-    audio_path = segment_dir / f"{source}.flac"
-    if audio_path.exists():
-        rel_path = f"{stream}/{segment_key}/{source}.flac"
-        audio_file = f"/app/speakers/api/serve_audio/{day}/{rel_path}"
+    audio_mimetype = None
+    audio_path = resolve_audio_file(segment_dir, source)
+    if audio_path is not None:
+        audio_file = audio_serve_url(day, stream, segment_key, audio_path.name)
+        audio_mimetype = MIME_TYPES[audio_path.suffix]
 
     parsed = segment_parse(segment_key)
     start_time, end_time = parsed if parsed[0] else (None, None)
@@ -992,6 +995,7 @@ def api_review(day: str, stream: str, segment_key: str, source: str) -> Any:
             "sentences": review_sentences,
             "all_entities": all_entities,
             "audio_file": audio_file,
+            "audio_mimetype": audio_mimetype,
             "has_labels": labels_data is not None,
             "summary": {
                 "total": len(review_sentences),
@@ -1937,4 +1941,7 @@ def serve_audio(day: str, rel_path: str) -> Any:
         return error
     if not path.is_file():
         return error_response(FILE_NOT_FOUND, detail="File not found")
-    return send_file(path, mimetype="audio/flac")
+    mimetype = MIME_TYPES.get(path.suffix.lower())
+    if mimetype is None:
+        raise ValueError(f"unregistered media extension for serve_audio: {path.suffix}")
+    return send_file(path, conditional=True, mimetype=mimetype)
diff --git a/solstone/apps/speakers/tests/conftest.py b/solstone/apps/speakers/tests/conftest.py
index 101d05aa0..24e5be8c1 100644
--- a/solstone/apps/speakers/tests/conftest.py
+++ b/solstone/apps/speakers/tests/conftest.py
@@ -92,6 +92,7 @@ def speakers_env(tmp_path, monkeypatch):
             *,
             stream: str | None = None,
             embeddings: np.ndarray | None = None,
+            audio_extension: str = ".flac",
         ) -> Path:
             """Create a segment with sentence embeddings.
 
@@ -114,7 +115,11 @@ def speakers_env(tmp_path, monkeypatch):
             )
 
             for source in sources:
-                lines = [json.dumps({"raw": f"{source}.flac", "model": "medium.en"})]
+                lines = [
+                    json.dumps(
+                        {"raw": f"{source}{audio_extension}", "model": "medium.en"}
+                    )
+                ]
 
                 # Parse segment_key to get base time (e.g., "143022_300" -> 14:30:22)
                 # This matches real transcriber output which uses absolute timestamps
@@ -159,7 +164,7 @@ def speakers_env(tmp_path, monkeypatch):
                         embeddings=source_embeddings,
                         statement_ids=statement_ids,
                     )
-                    (segment_dir / f"{source}.flac").write_bytes(b"")
+                    (segment_dir / f"{source}{audio_extension}").write_bytes(b"")
 
             return flat_dir
 
diff --git a/solstone/apps/speakers/tests/test_audio.py b/solstone/apps/speakers/tests/test_audio.py
new file mode 100644
index 000000000..7da660081
--- /dev/null
+++ b/solstone/apps/speakers/tests/test_audio.py
@@ -0,0 +1,74 @@
+# SPDX-License-Identifier: AGPL-3.0-only
+# Copyright (c) 2026 sol pbc
+
+"""Tests for speakers audio file resolution."""
+
+from __future__ import annotations
+
+from solstone.apps.speakers.audio import (
+    _ORDERED_AUDIO_EXTENSIONS,
+    audio_serve_url,
+    resolve_audio_file,
+    resolve_audio_url,
+)
+from solstone.think.media import AUDIO_EXTENSIONS
+
+
+def test_ordered_audio_extensions_match_media_registry() -> None:
+    assert frozenset(_ORDERED_AUDIO_EXTENSIONS) == AUDIO_EXTENSIONS
+    assert _ORDERED_AUDIO_EXTENSIONS[0] == ".flac"
+
+
+def test_resolve_audio_file_prefers_media_format_order(tmp_path) -> None:
+    opus_path = tmp_path / "mic_audio.opus"
+    flac_path = tmp_path / "mic_audio.flac"
+    opus_path.write_bytes(b"opus")
+    flac_path.write_bytes(b"flac")
+
+    assert resolve_audio_file(tmp_path, "mic_audio") == flac_path
+
+
+def test_resolve_audio_file_ignores_missing_and_unregistered_audio(tmp_path) -> None:
+    (tmp_path / "mic_audio.aac").write_bytes(b"aac")
+
+    assert resolve_audio_file(tmp_path, "mic_audio") is None
+
+
+def test_audio_serve_url_builds_expected_route() -> None:
+    assert (
+        audio_serve_url(
+            "20240101",
+            "test",
+            "143022_300",
+            "mic_audio.m4a",
+        )
+        == "/app/speakers/api/serve_audio/20240101/test/143022_300/mic_audio.m4a"
+    )
+
+
+def test_resolve_audio_url_uses_registered_audio_file(speakers_env) -> None:
+    env = speakers_env()
+    env.create_segment(
+        "20240101",
+        "143022_300",
+        ["mic_audio"],
+        audio_extension=".m4a",
+    )
+
+    assert (
+        resolve_audio_url(
+            "20240101",
+            "test",
+            "143022_300",
+            "mic_audio",
+        )
+        == "/app/speakers/api/serve_audio/20240101/test/143022_300/mic_audio.m4a"
+    )
+
+
+def test_resolve_audio_url_returns_none_without_registered_audio(speakers_env) -> None:
+    env = speakers_env()
+    _flat_dir, chronicle_dir = env._segment_dirs("20240101", "143022_300")
+    (chronicle_dir / "mic_audio.aac").write_bytes(b"aac")
+
+    assert resolve_audio_url("20240101", "test", "143022_300", "mic_audio") is None
diff --git a/solstone/apps/speakers/tests/test_discovery.py b/solstone/apps/speakers/tests/test_discovery.py
index 033aae05c..37a8eaed4 100644
--- a/solstone/apps/speakers/tests/test_discovery.py
+++ b/solstone/apps/speakers/tests/test_discovery.py
@@ -65,7 +65,12 @@ def _setup_owner_centroid(
     return centroid
 
 
-def _create_cluster_segments(env, embeddings: np.ndarray) -> list[tuple[str, str, int]]:
+def _create_cluster_segments(
+    env,
+    embeddings: np.ndarray,
+    *,
+    audio_extension: str = ".flac",
+) -> list[tuple[str, str, int]]:
     """Create four segments with one qualifying cluster and one filtered cluster."""
     segments = [
         ("20240101", "090000_300"),
@@ -79,7 +84,13 @@ def _create_cluster_segments(env, embeddings: np.ndarray) -> list[tuple[str, str
         segment_embeddings = embeddings
         if idx < 2:
             segment_embeddings = np.vstack([embeddings, alt_embeddings])
-        env.create_segment(day, segment_key, ["audio"], embeddings=segment_embeddings)
+        env.create_segment(
+            day,
+            segment_key,
+            ["audio"],
+            embeddings=segment_embeddings,
+            audio_extension=audio_extension,
+        )
         results.append((day, segment_key, segment_embeddings.shape[0]))
     return results
 
@@ -156,6 +167,38 @@ def test_discover_clusters_found(speakers_env):
     assert len(cluster["samples"]) == 3
 
 
+def test_discover_samples_use_registered_audio_extension(speakers_env):
+    env = speakers_env()
+    _setup_owner_centroid(env.journal, [0.0, 1.0])
+    embeddings = _make_speaker_embeddings([1.0, 0.0], 5)
+    _create_cluster_segments(env, embeddings, audio_extension=".m4a")
+
+    result = discover_unknown_speakers()
+
+    samples = result["clusters"][0]["samples"]
+    assert len(samples) == 3
+    for sample in samples:
+        assert sample["audio_url"] == (
+            f"/app/speakers/api/serve_audio/{sample['day']}/"
+            f"{sample['stream']}/{sample['segment_key']}/{sample['source']}.m4a"
+        )
+
+
+def test_discover_samples_allow_missing_audio(speakers_env):
+    env = speakers_env()
+    _setup_owner_centroid(env.journal, [0.0, 1.0])
+    embeddings = _make_speaker_embeddings([1.0, 0.0], 5)
+    segments = _create_cluster_segments(env, embeddings)
+    for day, segment_key, _sentence_count in segments:
+        (env.journal / "chronicle" / day / "test" / segment_key / "audio.flac").unlink()
+
+    result = discover_unknown_speakers()
+
+    samples = result["clusters"][0]["samples"]
+    assert len(samples) == 3
+    assert all(sample["audio_url"] is None for sample in samples)
+
+
 def test_discover_filters_attributed(speakers_env):
     env = speakers_env()
     _setup_owner_centroid(env.journal, [0.0, 1.0])
diff --git a/solstone/apps/speakers/tests/test_owner.py b/solstone/apps/speakers/tests/test_owner.py
index 48597b52e..121f1e585 100644
--- a/solstone/apps/speakers/tests/test_owner.py
+++ b/solstone/apps/speakers/tests/test_owner.py
@@ -452,6 +452,81 @@ def test_detect_owner_candidate_pool_ready(speakers_env):
     assert get_current()["voiceprint"]["status"] == "candidate"
 
 
+def test_owner_candidate_samples_use_registered_audio_extension(speakers_env):
+    from solstone.apps.speakers.owner import _owner_candidate_samples
+
+    env = speakers_env()
+    embeddings = _owner_embeddings(3, np.random.default_rng(1))
+    provenance = []
+    for idx, segment_key in enumerate(
+        ("090000_300", "091000_300", "092000_300"),
+        start=1,
+    ):
+        env.create_segment(
+            "20240101",
+            segment_key,
+            ["mic_audio"],
+            num_sentences=1,
+            embeddings=embeddings[idx - 1 : idx],
+            audio_extension=".m4a",
+        )
+        provenance.append(
+            {
+                "day": "20240101",
+                "stream": "test",
+                "segment_key": segment_key,
+                "source": "mic_audio",
+                "sentence_id": 1,
+                "duration_s": 5.0,
+            }
+        )
+
+    samples = _owner_candidate_samples(embeddings, embeddings[0], provenance)
+
+    assert len(samples) == 3
+    for sample in samples:
+        assert sample["audio_url"] == (
+            f"/app/speakers/api/serve_audio/{sample['day']}/"
+            f"{sample['stream']}/{sample['segment_key']}/{sample['source']}.m4a"
+        )
+
+
+def test_owner_candidate_samples_allow_missing_audio(speakers_env):
+    from solstone.apps.speakers.owner import _owner_candidate_samples
+
+    env = speakers_env()
+    embeddings = _owner_embeddings(1, np.random.default_rng(1))
+    env.create_segment(
+        "20240101",
+        "090000_300",
+        ["mic_audio"],
+        num_sentences=1,
+        embeddings=embeddings,
+    )
+    (
+        env.journal
+        / "chronicle"
+        / "20240101"
+        / "test"
+        / "090000_300"
+        / "mic_audio.flac"
+    ).unlink()
+    provenance = [
+        {
+            "day": "20240101",
+            "stream": "test",
+            "segment_key": "090000_300",
+            "source": "mic_audio",
+            "sentence_id": 1,
+            "duration_s": 5.0,
+        }
+    ]
+
+    samples = _owner_candidate_samples(embeddings, embeddings[0], provenance)
+
+    assert samples[0]["audio_url"] is None
+
+
 def test_detect_owner_candidate_selection_skips_rejected_and_non_principal(
     speakers_env,
 ):
diff --git a/solstone/apps/speakers/tests/test_routes.py b/solstone/apps/speakers/tests/test_routes.py
index 408ce5d97..8e356bf11 100644
--- a/solstone/apps/speakers/tests/test_routes.py
+++ b/solstone/apps/speakers/tests/test_routes.py
@@ -49,9 +49,11 @@ def serve_audio_client(tmp_path, monkeypatch):
     )
     segment_dir.mkdir(parents=True)
     (segment_dir / f"{SERVE_AUDIO_SOURCE}.flac").write_bytes(b"fLaC")
+    (segment_dir / f"{SERVE_AUDIO_SOURCE}.m4a").write_bytes(b"m4a")
     monkeypatch.setenv("SOLSTONE_JOURNAL", str(journal))
 
     app = create_app(str(journal))
+    app.config["TESTING"] = True
     return app.test_client(), journal
 
 
@@ -750,6 +752,41 @@ def test_serve_audio_sets_flac_mimetype(serve_audio_client):
     assert response.mimetype == "audio/flac"
 
 
+def test_serve_audio_sets_registered_mimetype(serve_audio_client):
+    """Serve audio endpoint returns the registered mimetype for sample playback."""
+    client, _journal = serve_audio_client
+
+    response = client.get(
+        f"/app/speakers/api/serve_audio/{SERVE_AUDIO_DAY}/"
+        f"{SERVE_AUDIO_STREAM}/{SERVE_AUDIO_SEGMENT}/{SERVE_AUDIO_SOURCE}.m4a"
+    )
+
+    assert response.status_code == 200
+    assert response.mimetype == "audio/mp4"
+
+
+def test_serve_audio_unregistered_extension_raises(serve_audio_client):
+    """Serve audio refuses existing files with unregistered extensions."""
+    client, journal = serve_audio_client
+    segment_dir = (
+        journal
+        / "chronicle"
+        / SERVE_AUDIO_DAY
+        / SERVE_AUDIO_STREAM
+        / SERVE_AUDIO_SEGMENT
+    )
+    (segment_dir / f"{SERVE_AUDIO_SOURCE}.aac").write_bytes(b"aac")
+
+    with pytest.raises(
+        ValueError,
+        match=r"unregistered media extension for serve_audio: \.aac",
+    ):
+        client.get(
+            f"/app/speakers/api/serve_audio/{SERVE_AUDIO_DAY}/"
+            f"{SERVE_AUDIO_STREAM}/{SERVE_AUDIO_SEGMENT}/{SERVE_AUDIO_SOURCE}.aac"
+        )
+
+
 def test_serve_audio_path_traversal_is_forbidden(serve_audio_client):
     """A rel_path resolving to a real file outside the day dir is refused 403."""
     client, journal = serve_audio_client
@@ -947,6 +984,10 @@ def test_api_review_with_labels(speakers_env):
         data = resp.get_json()
         assert data["has_labels"] is True
         assert data["summary"]["total"] > 0
+        assert data["audio_file"] == (
+            "/app/speakers/api/serve_audio/20240101/test/143022_300/mic_audio.flac"
+        )
+        assert data["audio_mimetype"] == "audio/flac"
         assert data["all_entities"][0]["name"] == "Alice Test"
         sentences = data["sentences"]
         s1 = next(s for s in sentences if s["id"] == 1)
@@ -975,6 +1016,61 @@ def test_api_review_no_labels(speakers_env):
         assert data["summary"]["needs_review"] == 0
 
 
+def test_api_review_uses_registered_audio_extension(speakers_env):
+    """Review endpoint reports the actual registered segment audio file."""
+    from flask import Flask
+
+    from solstone.apps.speakers.routes import speakers_bp
+
+    env = speakers_env()
+    env.create_segment(
+        "20240101",
+        "143022_300",
+        ["mic_audio"],
+        audio_extension=".m4a",
+    )
+
+    app = Flask(__name__)
+    app.register_blueprint(speakers_bp)
+
+    with app.test_client() as client:
+        resp = client.get("/app/speakers/api/review/20240101/test/143022_300/mic_audio")
+        assert resp.status_code == 200
+        data = resp.get_json()
+        assert data["audio_file"] == (
+            "/app/speakers/api/serve_audio/20240101/test/143022_300/mic_audio.m4a"
+        )
+        assert data["audio_mimetype"] == "audio/mp4"
+
+
+def test_api_review_omits_audio_when_registered_audio_is_missing(speakers_env):
+    """Review endpoint returns null audio metadata when source audio is purged."""
+    from flask import Flask
+
+    from solstone.apps.speakers.routes import speakers_bp
+
+    env = speakers_env()
+    env.create_segment("20240101", "143022_300", ["mic_audio"])
+    (
+        env.journal
+        / "chronicle"
+        / "20240101"
+        / "test"
+        / "143022_300"
+        / "mic_audio.flac"
+    ).unlink()
+
+    app = Flask(__name__)
+    app.register_blueprint(speakers_bp)
+
+    with app.test_client() as client:
+        resp = client.get("/app/speakers/api/review/20240101/test/143022_300/mic_audio")
+        assert resp.status_code == 200
+        data = resp.get_json()
+        assert data["audio_file"] is None
+        assert data["audio_mimetype"] is None
+
+
 def test_api_review_corrections_excludes_confirmed(speakers_env):
     """Corrections summary/filter state excludes user_confirmed labels."""
     import json
diff --git a/solstone/apps/speakers/workspace.html b/solstone/apps/speakers/workspace.html
index fd5f4b738..99934d424 100644
--- a/solstone/apps/speakers/workspace.html
+++ b/solstone/apps/speakers/workspace.html
@@ -1464,7 +1464,7 @@
               <div class="spk-owner-sample-meta">
                 ${escapeHtml(sample.day)} · ${escapeHtml(sample.stream)} · ${escapeHtml(sample.segment_key)}
               </div>
-              <audio controls preload="metadata" src="${sample.audio_url}"></audio>
+              ${sample.audio_url ? `<audio controls preload="metadata" src="${sample.audio_url}"></audio>` : ''}
             </div>
           `).join('')}
         </div>
@@ -1628,7 +1628,7 @@
               ${cluster.samples.map(sample => `
                 <div class="spk-discovery-sample">
                   <div class="spk-discovery-sample-text">${escapeHtml(sample.text || '(no transcript)')}</div>
-                  <audio controls preload="metadata" src="${sample.audio_url}"></audio>
+                  ${sample.audio_url ? `<audio controls preload="metadata" src="${sample.audio_url}"></audio>` : ''}
                 </div>
               `).join('')}
               <form class="spk-discovery-form">
@@ -2036,14 +2036,14 @@
   }
 
   function renderReview(data) {
-    const { audio_file } = data;
+    const { audio_file, audio_mimetype } = data;
     const audioContainer = document.getElementById('spkAudioContainer');
     if (audio_file) {
       audioContainer.innerHTML = `
         <div class="spk-player-controls">
           <button class="spk-skip-back" id="spkSkipBack" title="back 5s">&#8634; 5s</button>
           <audio id="spkAudio" controls preload="metadata">
-            <source src="${audio_file}" type="audio/flac">
+            <source src="${audio_file}" type="${audio_mimetype}">
           </audio>
           <button class="spk-speed-toggle" id="spkSpeedToggle" title="playback speed">1x</button>
         </div>
@@ -2823,7 +2823,7 @@
           <div class="spk-owner-sample">
             <div class="spk-card-title">sample ${index + 1}</div>
             <div class="spk-card-line">${escapeHtml(sample.day)} · ${escapeHtml(sample.stream)} · ${escapeHtml(sample.segment_key)}</div>
-            <audio controls preload="metadata" src="${escapeHtml(sample.audio_url)}"></audio>
+            ${sample.audio_url ? `<audio controls preload="metadata" src="${escapeHtml(sample.audio_url)}"></audio>` : ''}
           </div>
         `).join('')}
       </div>
diff --git a/tests/baselines/api/speakers/review.json b/tests/baselines/api/speakers/review.json
index 4715a319d..0d7eb29bb 100644
--- a/tests/baselines/api/speakers/review.json
+++ b/tests/baselines/api/speakers/review.json
@@ -167,6 +167,7 @@
     }
   ],
   "audio_file": null,
+  "audio_mimetype": null,
   "has_labels": true,
   "segment": {
     "end": "09:05",
-- 
2.51.2