fix(audio): add preferredFormat() fallback for WASAPI Float32 output devices by NF0T · Pull Request #3231 · aethersdr/AetherSDR

NF0T · 2026-05-27T23:52:53Z

Summary

ClientPuduMonitor::startPlayback() accepted only Int16 stereo at 24 kHz or 48 kHz. On Windows 11 with WASAPI shared mode, the audio engine mixer runs internally in Float32. When the shared pipeline is in Float32 mode, IAudioClient::IsFormatSupported() (surfaced via QAudioDevice::isFormatSupported()) rejects Int16 regardless of hardware capability — causing startPlayback() to bail at format negotiation on any Windows machine whose WASAPI mix format is not Int16.

Root cause confirmed: the same C-Media USB device opens successfully in AudioEngine as Float32 stereo 48 kHz — the hardware is not the constraint, the WASAPI shared pipeline is. This is the same class of format-negotiation problem already solved in QuindarLocalSink.cpp:61 and AudioEngine.cpp:306 via dev.preferredFormat().

Fix

When both Int16 @ 24 kHz and Int16 @ 48 kHz are rejected, query dev.preferredFormat(), enforce stereo channel count, and accept it if isFormatSupported() confirms it. After preparePlaybackPcm() produces the resampled Int16 buffer, convert Int16 → Float32 in-place when the negotiated sink format is Float32 (trivial per-sample multiply; the r8brain resampler already handles any rate conversion).

The accepted format and fallback reason are recorded in the AudioSummaryLogger open-success summary, making the path inspectable from future support bundles.

Files changed

File	Change
`src/core/ClientPuduMonitor.cpp`	Add `preferredFormat()` as third format candidate; add Int16→Float32 conversion post-resample when sink format is Float32

Test plan

Windows 11 with C-Media USB (or any WASAPI Float32 device): record audio → confirm actual audio playback through device, not just a recovered-but-silent RX stream
Confirm Audio Auxiliary sink summary log line now appears for "Aetherial monitor playback" with fallback reason showing preferredFormat path
Linux build: record and play back → confirm Int16 24 kHz fast path taken; preferredFormat branch unreached
macOS: confirm Int16 48 kHz path taken on standard CoreAudio devices
CI check-windows, check-macos, check-paths pass

Constitution

Principle VIII — Evidence Over Assertion: the WASAPI Float32 behavior is confirmed by the log in #3222 showing format=Float for AudioEngine's successful open of the same device that ClientPuduMonitor fails on. The fix follows QuindarLocalSink's established preferredFormat() pattern rather than asserting a novel approach.

Fixes #3229
Refs #3222

🤖 Generated with Claude Code

jensenpat · 2026-05-28T00:12:30Z

@AetherClaude does this touch the main audio sink or just the pudu recorder?

aethersdr-agent · 2026-05-28T00:13:22Z

@jensenpat Just the pudu recorder playback path — the change is scoped to ClientPuduMonitor::startPlayback() in src/core/ClientPuduMonitor.cpp (the local recap playback of the recorded snippet). The main RX audio sink in AudioEngine is not touched; it already has its own preferredFormat() handling at AudioEngine.cpp:306 (and QuindarLocalSink.cpp:61 for the Quindar tone sink). This PR brings ClientPuduMonitor in line with that established pattern rather than altering the main pipeline.

Concretely, the current code at ClientPuduMonitor.cpp:202-231 only tries Int16 stereo at 24 kHz then 48 kHz, then gives up — which is what fails when WASAPI shared mode is in Float32. The fix adds a third candidate via dev.preferredFormat() (forcing stereo), and when the negotiated format comes back as Float32 it converts the resampled Int16 buffer to Float32 in-place after preparePlaybackPcm(). No behaviour change on Linux (24 kHz Int16 fast path still wins) or macOS (48 kHz Int16 still wins on CoreAudio).

73, Jeremy KK7GWY & Claude (AI dev partner)

_{🤖 aethersdr-agent · cost: $1.5956 · model: claude-opus-4-7}

jensenpat · 2026-05-28T00:19:25Z

@AetherClaude should we move audio sink negotiation to it's own factory, so that we are not replicating sample rate issues in 3 places? what is the current number of sample rate issues and PRs? build a history table.

aethersdr-agent · 2026-05-28T00:21:04Z

@jensenpat Yes — a factory is the right call, and PR #3231 is the third instance that proves it. Grepping isFormatSupported / preferredFormat across src/core shows the same negotiation logic, with slightly different rate ladders and sample-format preferences, scattered across six sinks:

#	Sink	File:line	Format preference	preferredFormat() fallback?
1	RX engine / mic source	`AudioEngine.cpp:306,1125-1151,4263`	native rate first, then 48k/44k/24k	yes
2	Quindar local tone sink	`QuindarLocalSink.cpp:58-65`	Float 48k stereo	yes
3	CW sidetone	`CwSidetoneQAudioSink.cpp:53-59`	Float→Int16, multi-rate ladder	no (relies on Int16 fallback)
4	QSO recorder playback	`QsoRecorder.cpp:377-380`	Int16 24k → 48k, give up	no — same bug as Pudu pre-#3231
5	Pudu monitor playback	`ClientPuduMonitor.cpp:212-231`	Int16 24k → 48k, give up	adds it in #3231
6	Device diagnostics probe	`DeviceDiagnostics.cpp:63,102`	informational only	n/a

Sample-rate / format-negotiation PR + issue history

#	Title	State	Sink touched
#2629	(referenced in CwSidetone code) Int16-only VB-Audio Virtual Cable	closed	CW sidetone
#2668 / #2669	`fix(cw): Int16 fallback for sidetone sink on Int16-only WASAPI devices`	merged	CW sidetone
#2694	CWX sidetone latency / distortion on Windows FLEX-8400	closed	CW sidetone
#2295 / #2487	`fix(audio): replace QsoRecorder QTimer playback with QAudioSink pull mode`	merged	QSO recorder
#2403 / #2459	`fix(audio): restore L-R pan on Windows 48kHz resampler path`	merged	RX engine
#2929 / #3036	`fix(audio): WASAPI mono-only USB PnP mic silent-open recovery`	merged	mic source
#2930	`fix(audio): silent SSB/voice TX on macOS — honour mic native rate`	merged	mic source
#2971 / #2973	Default consolidated audio sink / sample-rate summary logging	merged	(cross-cutting log)
#3222	VUDU TX record/playback mutes rx	open	RX engine + Pudu
#3229	VUDU monitor playback silently fails on WASAPI Float32	open	Pudu
#3231	This PR — preferredFormat() fallback for Pudu	open	Pudu

That's 11 PRs/issues across 6 sinks, every one of them re-discovering some variant of "Qt's isFormatSupported() lies, and the negotiation rule depends on what the WASAPI shared-mode mixer happens to be in today." The next one to break is almost certainly QsoRecorder::startPlayback() at QsoRecorder.cpp:377-380 — it has the identical Int16-24k-then-48k-then-give-up shape that this PR is fixing in Pudu, and will fail the same way on the same WASAPI Float32 endpoint that motivated #3229.

Proposed factory shape

Something like AudioSinkNegotiator in src/core/audio/:

// src/core/audio/AudioSinkNegotiator.h
namespace aether::audio {

struct NegotiatedSink {
    QAudioFormat   format;              // actually accepted by the device
    int            actualRate = 0;      // == format.sampleRate()
    bool           fallbackOccurred = false;
    QStringList    fallbackReasons;     // human-readable, for AudioSummaryLogger
    QStringList    attemptedFormats;    // ordered probe trail
};

enum class FormatPreference {
    Int16First,   // QSO recorder, Pudu monitor: source PCM is Int16
    FloatFirst,   // Quindar, AudioEngine RX: source is Float32 already
    NativeFirst,  // CW sidetone: prefer whatever the device wants
};

struct NegotiationRequest {
    int  desiredRateHz = 48000;
    int  channels      = 2;
    FormatPreference preference = FormatPreference::Int16First;
    QList<int> rateCandidates;  // empty => {desired, 48000, 44100, 24000}
    QString    pathLabel;       // for the AudioSummaryLogger summary line
};

// Negotiates an output format, including preferredFormat() as the final
// candidate. Always logs the open-success/open-failure summary via
// AudioSummaryLogger so the path is inspectable from support bundles.
std::optional<NegotiatedSink>
negotiateOutput(const QAudioDevice& dev, const NegotiationRequest& req);

}  // namespace aether::audio

Callers shrink from ~25 lines of probe-ladder + summary-logger plumbing to:

auto sink = aether::audio::negotiateOutput(dev, {
    .desiredRateHz = kSampleRate,
    .channels      = kChannels,
    .preference    = aether::audio::FormatPreference::Int16First,
    .pathLabel     = QStringLiteral("Aetherial monitor playback"),
});
if (!sink) return;  // factory already logged the open-failure summary

…and the Int16→Float32 conversion-on-demand logic from #3231 belongs alongside it (a convertPcmToSinkFormat(QByteArray int16Pcm, const NegotiatedSink&) helper), so the next sink that needs it gets it for free.

Recommended sequencing

Land fix(audio): add preferredFormat() fallback for WASAPI Float32 output devices #3231 as-is — it's narrowly scoped, unblocks VUDU monitor playback silently fails on Windows WASAPI Float32 output devices #3229, and matches the established QuindarLocalSink pattern. Don't expand its scope.
Open a follow-up issue ("refactor: centralize audio sink format negotiation in AudioSinkNegotiator") that migrates all six call sites behind the factory, with QsoRecorder's missing-preferredFormat path as the motivating bug.
The follow-up touches files already on the check-windows / check-macos CI gates (added in CI: expand check-windows path filter to hot platform-guarded files #3052/ci: add AudioEngine to check-windows path filter (#3052) #3210/ci: add check-macos build gate for hot platform-guarded files #3223), so the refactor lands with platform coverage from day one.

The risk of doing it inside #3231 is that the refactor changes behaviour for CW sidetone and the main RX engine (different probe order, different log lines), and any regression there is a much louder failure than a silent Pudu monitor — keep the blast radius small.

A maintainer can authorize the orchestrator to land #3231's fix by adding the aetherclaude-eligible label (not currently on the issue). The factory refactor would be a separate, larger change worth scoping in its own issue first.

73, Jeremy KK7GWY & Claude (AI dev partner)

_{🤖 aethersdr-agent · cost: $4.1747 · model: claude-opus-4-7}

aethersdr-agent

Thanks for the fix, @NF0T — the precedent referenced (QuindarLocalSink.cpp:61 and the preferredFormat() use in AudioEngine.cpp:306) is the right pattern to follow here, and the root-cause analysis (WASAPI shared mode running a Float32 mix on a hardware device that itself accepts Int16) lines up with what the diagnostics in DeviceDiagnostics.cpp already track.

A few small things worth a second look before merge:

1. Sample-format scope of the preferredFormat branch (src/core/ClientPuduMonitor.cpp:235-249 in the new code)

The conversion block only handles QAudioFormat::Float:

if (fmt.sampleFormat() == QAudioFormat::Float) { /* int16 → float */ }

…but the acceptance check earlier accepts any format the device reports as supported:

if (preferred.isValid() && dev.isFormatSupported(preferred)) { fmt = preferred; ... }

The PR comment explicitly notes Float is the only format observed in practice on WASAPI/CoreAudio. That's a reasonable assumption — but the code doesn't enforce it. If a device's preferredFormat() ever returns Int32, UInt8, etc., the Int16 payload will be sent to a sink expecting a different layout and you'll get garbage rather than a clean failure.

Suggestion: gate the preferred acceptance on sampleFormat() == Int16 || sampleFormat() == Float (Int16 needs no conversion, Float uses the existing branch), and otherwise fall through to the existing failure log. Single extra condition, removes a latent footgun.

2. Interaction with #3230 (the bail/mute-lift fix)

The new preferredFormat-rejection failure path (the new else branch that logs "output device supports neither Int16 stereo nor its own preferredFormat") is a 7th early-return path. Once #3230 lands with its bail() helper, this branch will also need to call it or it'll re-introduce the stranded-mute bug on this specific (rare) failure mode. Worth flagging in whichever PR merges second so it doesn't slip.

3. Minor: summary.fallbackReason already covers it

The post-success summary will include "48000Hz Int16 unsupported -> preferredFormat (…)" in the fallback reason and the negotiated sampleFormat field on the summary — so the path is already inspectable from a support bundle as the test plan expects. Nothing to change, just confirming it works.

Otherwise the conversion math (/ 32768.0f) matches preparePlaybackPcm()'s existing scaling at ClientPuduMonitor.cpp:150-151, the buffer-size arithmetic is correct (frames * kChannels * sizeof(float)), and the scope is appropriately tight — single file, single function, mirrors prior art.

_{🤖 aethersdr-agent · cost: $6.9063 · model: claude-opus-4-7}

aethersdr-agent

Thanks @NF0T — the diagnosis (WASAPI shared mixer locking out Int16 despite hardware capability) is solid, and reusing the preferredFormat() pattern from QuindarLocalSink / AudioEngine is the right move. The Int16→Float32 conversion is straightforward and the m_playPcm = std::move(floatPcm) happens before m_playBuffer.setBuffer(&m_playPcm), so the QBuffer pointer is fine.

Two things worth tightening before merge:

1. Narrow what preferredFormat() is allowed to return

The post-resample conversion only handles QAudioFormat::Float:

if (fmt.sampleFormat() == QAudioFormat::Float) {
    // Int16 → Float32 conversion
}

…but the acceptance check above will accept anything isFormatSupported() confirms — Int32, UInt8, future formats, etc. If preferredFormat ever returns one of those (uncommon but not impossible on exotic CoreAudio aggregates or pro-audio drivers), the Int16 payload would be played through a sink expecting a different layout → garbage. The comment claims Float32 is the only realistic case, but the code doesn't enforce that claim. Suggest gating acceptance to Int16 or Float only:

const auto sf = preferred.sampleFormat();
const bool acceptable = (sf == QAudioFormat::Int16 || sf == QAudioFormat::Float);
if (preferred.isValid() && acceptable && dev.isFormatSupported(preferred)) {
    ...
}

That keeps the diagnostic path (logOpenFailure) honest for the rare unsupported case and prevents silent corruption.

2. Mute lifting on the new bail path

The new else branch logs the failure and returns without emitting muteRxRequested(false) — same class of bug as #3230. If #3230 lands first you'll need to wire the new path through its bail() lambda; if this lands first the same fix needs to be repeated here. Worth a coordinating note across the two PRs or rebasing to use the lambda once both are aligned.

Otherwise looks good — minor enough that I'd be happy to see it land once #2 in particular is resolved (either via #3230 or inline here).

_{🤖 aethersdr-agent · cost: $5.1402 · model: claude-opus-4-7}

jensenpat

Thanks for fix Ryan

…devices (aethersdr#3231) - Principle VIII. ClientPuduMonitor::startPlayback() accepted only Int16 stereo at 24 kHz or 48 kHz. On Windows WASAPI shared mode, Float32 mix pipelines can reject both Int16 candidates even when the device works through AudioEngine. When both Int16 candidates fail, query preferredFormat(), force stereo, and accept it if supported. Convert the prepared Int16 payload to Float32 when the sink negotiated Float32 so playback can start on WASAPI Float32 devices. The fallback is logged through AudioSummaryLogger, and the failure path preserves an inspectable reason if preferredFormat() is rejected. Fixes aethersdr#3229 Refs aethersdr#3222 Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com> Signed-off-by: jensenpat <patjensen@gmail.com>

…devices (aethersdr#3231) - Principle VIII. ClientPuduMonitor::startPlayback() accepted only Int16 stereo at 24 kHz or 48 kHz. On Windows WASAPI shared mode, Float32 mix pipelines can reject both Int16 candidates even when the device works through AudioEngine. When both Int16 candidates fail, query preferredFormat(), force stereo, and accept it if supported. Convert the prepared Int16 payload to Float32 when the sink negotiated Float32 so playback can start on WASAPI Float32 devices. This keeps the existing bail() paths from PR 3230 so RX mute is still lifted on every playback-start failure. The fallback is logged through AudioSummaryLogger, and the failure path preserves an inspectable reason if preferredFormat() is rejected. Fixes aethersdr#3229 Refs aethersdr#3222 Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com> Signed-off-by: jensenpat <patjensen@gmail.com>

jensenpat

Approved refreshed maintainer-signed squash. The conflict with #3230 is resolved by preserving the mute-release bail() paths while adding the preferredFormat()/Float32 fallback from this PR.

@jensenpat

…3397) ## What First, **pure-addition** step of the consolidated audio sink factory (#3306): a single shared, OS-parameterized format/rate **negotiation policy** that every audio sink and source will migrate onto, replacing the ~9 divergent per-sink fallback ladders and per-OS `#ifdef` branches that are the root of a recurring class of platform audio bugs. Zero behaviour change — this is the maintainer-endorsed *"land the helper first, migrate sinks incrementally"* approach (#3306, and the #3194 thread's "sensitive audio-stack changes must be separate PRs with soak time"). ## Why this shape The historical bugs escaped CI because the per-OS rate ladders were compiled behind `Q_OS_*`, so a Linux runner could only ever exercise the Linux path. The new policy is a **pure function over an injected `DeviceCaps` snapshot with `TargetOs` as a parameter, not an `#ifdef`** — so one headless, hardware-free test binary exercises the Windows, macOS *and* Linux ladders on any runner. ## Contents - **`src/core/AudioFormatNegotiator.{h,cpp}`** — dependency-free policy (links only `Qt6::Core`). One ladder owns: - Windows force-48k + r8brain (#2120/#2123), macOS 48k/A2DP (#1705), Linux native-24k — divergences preserved as **data**, not `#ifdef`. - The **universal 44.1 kHz rung** that RX/Quindar/QSO are missing today (#3385). - `Int16`↔`Float32` + `preferredFormat()` catch-all (#2669 / #1090 / #3231). - macOS mic preferred-rate-first (#2930) and Bluetooth-HFP native rate (#2615); Windows probe-at-open (#2929). - The two stereo resampler strategies — **PreservePan** (RX/QSO) vs **MonoCollapse** (TCI-TX/RADE) — kept deliberately distinct (#2403/#2459). - **`tests/audio_format_negotiation_test.cpp`** — table-driven golden matrix, registered under CTest. **25/25 pass**, including the 44.1k-only-device regression guard. - **`docs/audio-sink-factory.md`** — the full three-layer design (pure policy → live Qt wrapper → device-ownership router), the device-following **"uncoupling"** fix (CW / RADE / Aetherial-Audio "Pudu" recorder following the selected output), the regression-guard invariants, and the one-sink-per-PR migration plan. ## Test ``` $ ./build/audio_format_negotiation_test ... 25/25 checks passed ``` ## Follow-ups (per the migration plan in the doc) Live Qt wrapper → migrate RX speaker → migrate PC mic → `AudioOutputRouter` (closes the uncoupling class) → TCI-TX client-rate → PipeWire DAX shared resampler. Each a separate, soakable PR. Refs #3306 💻 Generated with Claude Code (Opus 4.8) with architecture by @jensenpat Co-authored-by: Codex <noreply@openai.com>

…ersdr#3306) Phase 6b enabler. The Output ladder is Float-first by default (right for the float-native RX speaker / sidetone / Quindar). The Int16-native playback sinks (QSO playback, Pudu monitor — they hold recorded/captured int16) want the opposite: Int16 on a normal device so they need no conversion, with Float only as the fallback for Float-only endpoints (aethersdr#3231). Add an optional `FormatPreference { Auto, Int16First, Float32First }` threaded through buildLadder()/negotiate() and the AudioDeviceNegotiator wrapper. Default Auto preserves today's behaviour exactly (RX/mic untouched). Three golden rows document the Int16-first output: normal device -> Int16 at the per-OS rate (Win/Mac 48k, Linux native 24k), Float-only device -> Float fallback. 28/28. Refs aethersdr#3306 Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

aethersdr#3306 6b) Both playback sinks held their own bespoke Int16 rate ladders. They now negotiate via AudioDeviceNegotiator with the new Int16First preference: on a normal device they get Int16 (no conversion, as before) at the per-OS rate (Win/Mac 48k — dodging the WASAPI 24k resampler artifacts aethersdr#2120, same policy as the RX sink — Linux native 24k), with the universal 44.1k and preferredFormat fallbacks supplied in one place. They walk the ladder with isFormatSupported (trusted — that is exactly how aethersdr#3231's Float-only WASAPI devices are detected). ClientPuduMonitor keeps its Int16->Float conversion (aethersdr#3231). QsoRecorder playback *gains* the same guard: it previously bailed outright on a Float-only device; now it falls back to Float and converts. Net behaviour on normal devices is unchanged except the 24k->48k preferred-rate shift on Win/Mac (consistent with RX; the recording is resampled up-front, one-shot). Builds & links clean; negotiation 28/28, wrapper 8/8, router 9/9. Refs aethersdr#3306 Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

@jensenpat

…3306 6b/6c) (#3655) > **Stacks on #3631 → #3630.** First two commits are #3630 (router) + #3631 (Quindar); the new commits are the enabler + 6b + 6c. Merge the parents first and this collapses to the three new commits. Each is a clean standalone commit. Closes out **Phase 6** of the audio sink factory (#3306) — every auxiliary output sink now negotiates format through the shared factory instead of a bespoke per-sink ladder. ## Commits 1. **Int16-first `FormatPreference` enabler** — the Output ladder is Float-first (right for the float-native RX/sidetone/Quindar). The Int16-native playback sinks want the opposite, so this adds an optional `FormatPreference { Auto, Int16First, Float32First }` threaded through `buildLadder`/`negotiate` + the wrapper. Default `Auto` leaves RX/mic **byte-identical**. +3 golden rows (28/28). 2. **6b — Pudu monitor + QSO playback** negotiate via `AudioDeviceNegotiator(Int16First, PreservePan)`. On a normal device they get **Int16 at the per-OS rate** (no conversion, as before) with the universal 44.1k + preferredFormat fallbacks in one place. `ClientPuduMonitor` keeps its Int16→Float guard (#3231); **`QsoRecorder` playback gains the same guard** — it previously *bailed* on a Float-only WASAPI device, now it works. 3. **6c — CW sidetone (QAudio backend)** walks the Float-first ladder (Int16 fallback preserved for VB-Audio #2629). The **PortAudio backend is untouched** (Qt negotiator can't drive it). ## Behaviour notes (for soak) Normal devices are unchanged **except** Pudu/QSO/CW now prefer **48k on Win/Mac** (was 24k for Pudu/QSO) — this is the factory's deliberate #2120 policy (dodges WASAPI 24k-resampler artifacts, same as the RX sink); playback is resampled up-front, one-shot. Linux stays native 24k. Strictly-additive wins: QSO playback now survives Float-only devices; all three gain the 44.1k rung. ## Test / build - `audio_format_negotiation_test` **28/28**, `audio_device_negotiator_test` 8/8, `audio_output_router_test` 9/9 — all green under CTest. - Full macOS app builds & links clean (RelWithDebInfo); deployed to the macOS test box. After this, the factory covers every device-facing sink. Remaining #3306 items (TCI-TX client-rate, PipeWire DAX resampler, Windows mic) are separate. Refs #3306 💻 Generated with Claude Code (Opus 4.8) with architecture by @jensenpat --------- Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>

…devices (aethersdr#3231) - Principle VIII. ClientPuduMonitor::startPlayback() accepted only Int16 stereo at 24 kHz or 48 kHz. On Windows WASAPI shared mode, Float32 mix pipelines can reject both Int16 candidates even when the device works through AudioEngine. When both Int16 candidates fail, query preferredFormat(), force stereo, and accept it if supported. Convert the prepared Int16 payload to Float32 when the sink negotiated Float32 so playback can start on WASAPI Float32 devices. This keeps the existing bail() paths from PR 3230 so RX mute is still lifted on every playback-start failure. The fallback is logged through AudioSummaryLogger, and the failure path preserves an inspectable reason if preferredFormat() is rejected. Fixes aethersdr#3229 Refs aethersdr#3222 Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com> Signed-off-by: jensenpat <patjensen@gmail.com>

@jensenpat

…#3306) (aethersdr#3397) ## What First, **pure-addition** step of the consolidated audio sink factory (aethersdr#3306): a single shared, OS-parameterized format/rate **negotiation policy** that every audio sink and source will migrate onto, replacing the ~9 divergent per-sink fallback ladders and per-OS `#ifdef` branches that are the root of a recurring class of platform audio bugs. Zero behaviour change — this is the maintainer-endorsed *"land the helper first, migrate sinks incrementally"* approach (aethersdr#3306, and the aethersdr#3194 thread's "sensitive audio-stack changes must be separate PRs with soak time"). ## Why this shape The historical bugs escaped CI because the per-OS rate ladders were compiled behind `Q_OS_*`, so a Linux runner could only ever exercise the Linux path. The new policy is a **pure function over an injected `DeviceCaps` snapshot with `TargetOs` as a parameter, not an `#ifdef`** — so one headless, hardware-free test binary exercises the Windows, macOS *and* Linux ladders on any runner. ## Contents - **`src/core/AudioFormatNegotiator.{h,cpp}`** — dependency-free policy (links only `Qt6::Core`). One ladder owns: - Windows force-48k + r8brain (aethersdr#2120/aethersdr#2123), macOS 48k/A2DP (aethersdr#1705), Linux native-24k — divergences preserved as **data**, not `#ifdef`. - The **universal 44.1 kHz rung** that RX/Quindar/QSO are missing today (aethersdr#3385). - `Int16`↔`Float32` + `preferredFormat()` catch-all (aethersdr#2669 / aethersdr#1090 / aethersdr#3231). - macOS mic preferred-rate-first (aethersdr#2930) and Bluetooth-HFP native rate (aethersdr#2615); Windows probe-at-open (aethersdr#2929). - The two stereo resampler strategies — **PreservePan** (RX/QSO) vs **MonoCollapse** (TCI-TX/RADE) — kept deliberately distinct (aethersdr#2403/aethersdr#2459). - **`tests/audio_format_negotiation_test.cpp`** — table-driven golden matrix, registered under CTest. **25/25 pass**, including the 44.1k-only-device regression guard. - **`docs/audio-sink-factory.md`** — the full three-layer design (pure policy → live Qt wrapper → device-ownership router), the device-following **"uncoupling"** fix (CW / RADE / Aetherial-Audio "Pudu" recorder following the selected output), the regression-guard invariants, and the one-sink-per-PR migration plan. ## Test ``` $ ./build/audio_format_negotiation_test ... 25/25 checks passed ``` ## Follow-ups (per the migration plan in the doc) Live Qt wrapper → migrate RX speaker → migrate PC mic → `AudioOutputRouter` (closes the uncoupling class) → TCI-TX client-rate → PipeWire DAX shared resampler. Each a separate, soakable PR. Refs aethersdr#3306 💻 Generated with Claude Code (Opus 4.8) with architecture by @jensenpat Co-authored-by: Codex <noreply@openai.com>

NF0T requested a review from a team as a code owner May 27, 2026 23:52

NF0T mentioned this pull request May 28, 2026

VUDU TX record / playback mutes rx #3222

Closed

2 tasks

aethersdr-agent Bot reviewed May 28, 2026

View reviewed changes

jensenpat previously approved these changes May 28, 2026

View reviewed changes

jensenpat self-assigned this May 28, 2026

jensenpat force-pushed the fix/vudu-wasapi-float-fallback branch from a25aae5 to 7de3d5e Compare May 28, 2026 05:06

jensenpat dismissed their stale review via aaf85cf May 28, 2026 05:08

jensenpat force-pushed the fix/vudu-wasapi-float-fallback branch from 7de3d5e to aaf85cf Compare May 28, 2026 05:08

jensenpat approved these changes May 28, 2026

View reviewed changes

jensenpat merged commit 51f275d into aethersdr:main May 28, 2026
6 checks passed

NF0T deleted the fix/vudu-wasapi-float-fallback branch May 29, 2026 15:41

This was referenced May 29, 2026

AetherSDR audio muting issue during mic setup recording #3278

Closed

Aethervoice TX record / playback not working on Windows 11 version 26.6.1.1 #3361

Closed

jensenpat mentioned this pull request Jun 5, 2026

feat(audio): consolidated sink/rate negotiation foundation (#3306) #3397

Merged

jensenpat mentioned this pull request Jun 19, 2026

refactor(audio): finish Phase 6 — Pudu/QSO/CW sinks onto the factory (#3306 6b/6c) #3655

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

fix(audio): add preferredFormat() fallback for WASAPI Float32 output devices#3231

fix(audio): add preferredFormat() fallback for WASAPI Float32 output devices#3231
jensenpat merged 1 commit into
aethersdr:mainfrom
NF0T:fix/vudu-wasapi-float-fallback

NF0T commented May 27, 2026

Uh oh!

jensenpat commented May 28, 2026

Uh oh!

aethersdr-agent Bot commented May 28, 2026

Uh oh!

jensenpat commented May 28, 2026

Uh oh!

aethersdr-agent Bot commented May 28, 2026

Uh oh!

aethersdr-agent Bot left a comment

Uh oh!

aethersdr-agent Bot left a comment

Uh oh!

jensenpat left a comment

Uh oh!

jensenpat left a comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

NF0T commented May 27, 2026

Summary

Fix

Files changed

Test plan

Constitution

Uh oh!

jensenpat commented May 28, 2026

Uh oh!

aethersdr-agent Bot commented May 28, 2026

Uh oh!

jensenpat commented May 28, 2026

Uh oh!

aethersdr-agent Bot commented May 28, 2026

Sample-rate / format-negotiation PR + issue history

Proposed factory shape

Recommended sequencing

Uh oh!

aethersdr-agent Bot left a comment

Choose a reason for hiding this comment

Uh oh!

aethersdr-agent Bot left a comment

Choose a reason for hiding this comment

Uh oh!

jensenpat left a comment

Choose a reason for hiding this comment

Uh oh!

jensenpat left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants