Merge pull request #8 from recurse-chat/fix-linux-audio-and-crypto

BUG: Fix linux audio and crypto

src-tauri/Cargo.lock

@@ -73,6 +73,22 @@ dependencies = [
  "libc",
 ]
 
+[[package]]
+name = "annotate-snippets"
+version = "0.11.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "710e8eae58854cdc1790fcb56cca04d712a17be849eeb81da2a724bf4bae2bc4"
+dependencies = [
+ "anstyle",
+ "unicode-width",
+]
+
+[[package]]
+name = "anstyle"
+version = "1.0.14"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "940b3a0ca603d1eade50a4846a2afffd5ef57a9feac2c0e2ec2e14f9ead76000"
+
 [[package]]
 name = "anyhow"
 version = "1.0.104"
@@ -230,7 +246,7 @@ dependencies = [
  "glib-sys",
  "gobject-sys",
  "libc",
- "system-deps",
+ "system-deps 6.2.2",
 ]
 
 [[package]]
@@ -290,6 +306,25 @@ version = "0.22.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "72b3254f16251a8381aa12e40e3c4d2f0199f8c6508fbecb9d91f575e0fbb8c6"
 
+[[package]]
+name = "bindgen"
+version = "0.72.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "993776b509cfb49c750f11b8f07a46fa23e0a1386ffc01fb1e7d343efc387895"
+dependencies = [
+ "annotate-snippets",
+ "bitflags 2.13.1",
+ "cexpr",
+ "clang-sys",
+ "itertools",
+ "proc-macro2",
+ "quote",
+ "regex",
+ "rustc-hash",
+ "shlex 1.3.0",
+ "syn 2.0.119",
+]
+
 [[package]]
 name = "bit-set"
 version = "0.8.0"
@@ -439,7 +474,7 @@ checksum = "685c9fa8e590b8b3d678873528d83411db17242a73fccaed827770ea0fedda51"
 dependencies = [
  "glib-sys",
  "libc",
- "system-deps",
+ "system-deps 6.2.2",
 ]
 
 [[package]]
@@ -491,7 +526,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "509591b7bcd67f4ef775afad7662703b4935daaa6ec0e5605cfb1090b32a2b6d"
 dependencies = [
  "find-msvc-tools",
- "shlex",
+ "shlex 2.0.1",
 ]
 
 [[package]]
@@ -500,6 +535,15 @@ version = "1.1.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "6d43a04d8753f35258c91f8ec639f792891f748a1edbd759cf1dcea3382ad83c"
 
+[[package]]
+name = "cexpr"
+version = "0.6.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "6fac387a98bb7c37292057cffc56d62ecb629900026402633ae9160df93a8766"
+dependencies = [
+ "nom 7.1.3",
+]
+
 [[package]]
 name = "cfb"
 version = "0.7.3"
@@ -518,7 +562,17 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "d067ad48b8650848b989a59a86c6c36a995d02d2bf778d45c3c5d57bc2718f02"
 dependencies = [
  "smallvec",
- "target-lexicon",
+ "target-lexicon 0.12.16",
+]
+
+[[package]]
+name = "cfg-expr"
+version = "0.20.9"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "fe4ece8474b5f766c63426647e7b4b316b67431ade1036a8313cee24a03ae917"
+dependencies = [
+ "smallvec",
+ "target-lexicon 0.13.5",
 ]
 
 [[package]]
@@ -573,6 +627,17 @@ dependencies = [
  "inout",
 ]
 
+[[package]]
+name = "clang-sys"
+version = "1.9.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "157a8ba7b480713b56f4c09fd13fc3e0a22a5dfab8097ba61cbc5feef950788a"
+dependencies = [
+ "glob",
+ "libc",
+ "libloading 0.8.9",
+]
+
 [[package]]
 name = "cmov"
 version = "0.5.4"
@@ -608,6 +673,12 @@ dependencies = [
  "version_check",
 ]
 
+[[package]]
+name = "cookie-factory"
+version = "0.3.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9885fa71e26b8ab7855e2ec7cae6e9b380edff76cd052e07c683a0319d51b3a2"
+
 [[package]]
 name = "core-foundation"
 version = "0.10.1"
@@ -672,6 +743,7 @@ dependencies = [
  "block2",
  "coreaudio-rs",
  "dasp_sample",
+ "futures",
  "jni 0.22.4",
  "js-sys",
  "libc",
@@ -687,6 +759,9 @@ dependencies = [
  "objc2-core-audio-types",
  "objc2-core-foundation",
  "objc2-foundation",
+ "pipewire",
+ "portable-atomic",
+ "pulseaudio",
  "web-sys",
  "windows 0.62.2",
  "windows-core 0.62.2",
@@ -1058,6 +1133,12 @@ version = "1.0.20"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "d0881ea181b1df73ff77ffaaf9c7544ecc11e82fba9b5f27b262a3c73a332555"
 
+[[package]]
+name = "either"
+version = "1.16.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "91622ff5e7162018101f2fea40d6ebf4a78bbe5a49736a2020649edf9693679e"
+
 [[package]]
 name = "embed-resource"
 version = "3.0.11"
@@ -1099,6 +1180,17 @@ version = "1.1.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "66b7e2430c6dff6a955451e2cfc438f09cea1965a9d6f87f7e3b90decc014099"
 
+[[package]]
+name = "enum-primitive-derive"
+version = "0.3.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ba7795da175654fe16979af73f81f26a8ea27638d8d9823d317016888a63dc4c"
+dependencies = [
+ "num-traits",
+ "quote",
+ "syn 2.0.119",
+]
+
 [[package]]
 name = "enumflags2"
 version = "0.7.12"
@@ -1256,6 +1348,21 @@ dependencies = [
  "percent-encoding",
 ]
 
+[[package]]
+name = "futures"
+version = "0.3.32"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "8b147ee9d1f6d097cef9ce628cd2ee62288d963e16fb287bd9286455b241382d"
+dependencies = [
+ "futures-channel",
+ "futures-core",
+ "futures-executor",
+ "futures-io",
+ "futures-sink",
+ "futures-task",
+ "futures-util",
+]
+
 [[package]]
 name = "futures-channel"
 version = "0.3.34"
@@ -1263,6 +1370,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "b1f9e3d69d39e4862ffed03ed071a76f9a13ba1d9109d355b0f0aa6b15e393c4"
 dependencies = [
  "futures-core",
+ "futures-sink",
 ]
 
 [[package]]
@@ -1330,6 +1438,7 @@ version = "0.3.34"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "0d50a92467f8ba5dd6e3ee5d4bd04d73ab2e4e1c44474a0674821dfce14b79bc"
 dependencies = [
+ "futures-channel",
  "futures-core",
  "futures-io",
  "futures-macro",
@@ -1378,7 +1487,7 @@ dependencies = [
  "glib-sys",
  "gobject-sys",
  "libc",
- "system-deps",
+ "system-deps 6.2.2",
 ]
 
 [[package]]
@@ -1395,7 +1504,7 @@ dependencies = [
  "libc",
  "pango-sys",
  "pkg-config",
- "system-deps",
+ "system-deps 6.2.2",
 ]
 
 [[package]]
@@ -1409,7 +1518,7 @@ dependencies = [
  "gobject-sys",
  "libc",
  "pkg-config",
- "system-deps",
+ "system-deps 6.2.2",
 ]
 
 [[package]]
@@ -1435,7 +1544,7 @@ dependencies = [
  "gdk-sys",
  "glib-sys",
  "libc",
- "system-deps",
+ "system-deps 6.2.2",
  "x11",
 ]
 
@@ -1512,7 +1621,7 @@ dependencies = [
  "glib-sys",
  "gobject-sys",
  "libc",
- "system-deps",
+ "system-deps 6.2.2",
  "winapi",
 ]
 
@@ -1560,7 +1669,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "063ce2eb6a8d0ea93d2bf8ba1957e78dbab6be1c2220dd3daca57d5a9d869898"
 dependencies = [
  "libc",
- "system-deps",
+ "system-deps 6.2.2",
 ]
 
 [[package]]
@@ -1577,7 +1686,7 @@ checksum = "0850127b514d1c4a4654ead6dedadb18198999985908e6ffe4436f53c785ce44"
 dependencies = [
  "glib-sys",
  "libc",
- "system-deps",
+ "system-deps 6.2.2",
 ]
 
 [[package]]
@@ -1616,7 +1725,7 @@ dependencies = [
  "gobject-sys",
  "libc",
  "pango-sys",
- "system-deps",
+ "system-deps 6.2.2",
 ]
 
 [[package]]
@@ -1979,6 +2088,15 @@ dependencies = [
  "once_cell",
 ]
 
+[[package]]
+name = "itertools"
+version = "0.13.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "413ee7dfc52ee1a4949ceeb7dbc8a33f2d6c088194d9f922fb8318faf1f01186"
+dependencies = [
+ "either",
+]
+
 [[package]]
 name = "itoa"
 version = "1.0.18"
@@ -2005,7 +2123,7 @@ dependencies = [
  "glib-sys",
  "gobject-sys",
  "libc",
- "system-deps",
+ "system-deps 6.2.2",
 ]
 
 [[package]]
@@ -2199,7 +2317,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "6e9ec52138abedcc58dc17a7c6c0c00a2bdb4f3427c7f63fa97fd0d859155caf"
 dependencies = [
  "gtk-sys",
- "libloading",
+ "libloading 0.7.4",
  "once_cell",
 ]
 
@@ -2228,6 +2346,16 @@ dependencies = [
  "winapi",
 ]
 
+[[package]]
+name = "libloading"
+version = "0.8.9"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d7c4b02199fee7c5d21a5ae7d8cfa79a6ef5bb2fc834d6e9058e89c825efdc55"
+dependencies = [
+ "cfg-if",
+ "windows-link 0.2.1",
+]
+
 [[package]]
 name = "libredox"
 version = "0.1.20"
@@ -2237,6 +2365,33 @@ dependencies = [
  "libc",
 ]
 
+[[package]]
+name = "libspa"
+version = "0.10.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "882f7427e7989dcc9d388b7f05c4630390a1d7696f9ffa469cd4a7a48f0b4c40"
+dependencies = [
+ "bitflags 2.13.1",
+ "cc",
+ "cookie-factory",
+ "libc",
+ "libspa-sys",
+ "nom 8.0.0",
+ "rustix",
+ "system-deps 7.0.8",
+]
+
+[[package]]
+name = "libspa-sys"
+version = "0.10.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "3b6e17bdaf63ed0d5e4144022624032b41fd9733112e8c74ac26fc9bf1291924"
+dependencies = [
+ "bindgen",
+ "cc",
+ "system-deps 7.0.8",
+]
+
 [[package]]
 name = "linux-raw-sys"
 version = "0.12.1"
@@ -2302,6 +2457,12 @@ version = "0.3.17"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "6877bb514081ee2a7ff5ef9de3281f14a4dd4bceac4c09388074a6b5df8a139a"
 
+[[package]]
+name = "minimal-lexical"
+version = "0.2.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "68354c5c6bd36d73ff3feceb05efa59b6acb7626617f4962be322a825e61f79a"
+
 [[package]]
 name = "miniz_oxide"
 version = "0.8.9"
@@ -2319,6 +2480,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "30d65c71f1ce40ab09135ce117d742b9f8a19ff91a41a8b57ed50bc2de59c427"
 dependencies = [
  "libc",
+ "log",
  "wasi",
  "windows-sys 0.61.2",
 ]
@@ -2380,6 +2542,25 @@ version = "1.0.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "650eef8c711430f1a879fdd01d4745a7deea475becfb90269c06775983bbf086"
 
+[[package]]
+name = "nom"
+version = "7.1.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d273983c5a657a70a3e8f2a01329822f3b8c8172b73826411a55751e404a0a4a"
+dependencies = [
+ "memchr",
+ "minimal-lexical",
+]
+
+[[package]]
+name = "nom"
+version = "8.0.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "df9761775871bdef83bee530e60050f7e54b1105350d6884eb0fb4f46c2f9405"
+dependencies = [
+ "memchr",
+]
+
 [[package]]
 name = "num-complex"
 version = "0.4.6"
@@ -2748,7 +2929,7 @@ dependencies = [
  "glib-sys",
  "gobject-sys",
  "libc",
- "system-deps",
+ "system-deps 6.2.2",
 ]
 
 [[package]]
@@ -2856,6 +3037,31 @@ dependencies = [
  "futures-io",
 ]
 
+[[package]]
+name = "pipewire"
+version = "0.10.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "bde71084c4e25959d68f1ea54daa75e5ecdb338e5caf0b5510143b79baa32d5c"
+dependencies = [
+ "bitflags 2.13.1",
+ "libc",
+ "libspa",
+ "libspa-sys",
+ "pipewire-sys",
+ "rustix",
+]
+
+[[package]]
+name = "pipewire-sys"
+version = "0.10.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9ce653f53e63e5b93853218092ee9a8906a5d082c92f3f1db26316955dd63ce0"
+dependencies = [
+ "bindgen",
+ "libspa-sys",
+ "system-deps 7.0.8",
+]
+
 [[package]]
 name = "pkg-config"
 version = "0.3.34"
@@ -3032,6 +3238,22 @@ dependencies = [
  "unicode-ident",
 ]
 
+[[package]]
+name = "pulseaudio"
+version = "0.3.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d70623bd7967a9ca4c2ae0e807fc380b291f98480fc037042305ec643a4d3373"
+dependencies = [
+ "bitflags 2.13.1",
+ "byteorder",
+ "enum-primitive-derive",
+ "futures",
+ "log",
+ "mio",
+ "num-traits",
+ "thiserror 1.0.69",
+]
+
 [[package]]
 name = "quick-xml"
 version = "0.41.0"
@@ -3526,6 +3748,12 @@ dependencies = [
  "digest",
 ]
 
+[[package]]
+name = "shlex"
+version = "1.3.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "0fda2ff0d084019ba4d7c6f371c95d8fd75ce3524c3cb8fb653a3023f6323e64"
+
 [[package]]
 name = "shlex"
 version = "2.0.1"
@@ -3637,7 +3865,7 @@ dependencies = [
  "glib-sys",
  "gobject-sys",
  "libc",
- "system-deps",
+ "system-deps 6.2.2",
 ]
 
 [[package]]
@@ -3751,13 +3979,26 @@ version = "6.2.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "a3e535eb8dded36d55ec13eddacd30dec501792ff23a0b1682c38601b8cf2349"
 dependencies = [
- "cfg-expr",
+ "cfg-expr 0.15.8",
  "heck 0.5.0",
  "pkg-config",
  "toml 0.8.2",
  "version-compare",
 ]
 
+[[package]]
+name = "system-deps"
+version = "7.0.8"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "396a35feb67335377e0251fcbc1092fc85c484bd4e3a7a54319399da127796e7"
+dependencies = [
+ "cfg-expr 0.20.9",
+ "heck 0.5.0",
+ "pkg-config",
+ "toml 1.1.4+spec-1.1.0",
+ "version-compare",
+]
+
 [[package]]
 name = "tao"
 version = "0.35.3"
@@ -3815,6 +4056,12 @@ version = "0.12.16"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "61c41af27dd6d1e27b1b16b489db798443478cef1f06a660c96db617ba5de3b1"
 
+[[package]]
+name = "target-lexicon"
+version = "0.13.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "adb6935a6f5c20170eeceb1a3835a49e12e19d792f6dd344ccc76a985ca5a6ca"
+
 [[package]]
 name = "tauri"
 version = "2.11.5"
@@ -4520,6 +4767,12 @@ version = "1.13.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "c6f5d3c3b1bf09027a88a6bc961fc00497d651009560b5463668dc81b0fa87a8"
 
+[[package]]
+name = "unicode-width"
+version = "0.2.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b4ac048d71ede7ee76d585517add45da530660ef4390e49b098733c6e897f254"
+
 [[package]]
 name = "universal-hash"
 version = "0.6.1"
@@ -4781,7 +5034,7 @@ dependencies = [
  "libc",
  "pkg-config",
  "soup3-sys",
- "system-deps",
+ "system-deps 6.2.2",
 ]
 
 [[package]]

src-tauri/Cargo.toml

@@ -22,7 +22,7 @@ tauri = { version = "2", features = [] }
 tauri-plugin-opener = "2"
 serde = { version = "1", features = ["derive"] }
 serde_json = "1"
-cpal = "0.18.2"
+cpal = { version = "0.18.2", features = ["pipewire", "pulseaudio"] }
 ringbuf = "0.5.1"
 rubato = "5.0.0"
 chacha20poly1305 = "0.11.0"

src-tauri/src/commands/audio.rs

@@ -10,7 +10,10 @@ use std::{
 };
 
 use chacha20poly1305::{aead::KeyInit, ChaCha20Poly1305, Key};
-use cpal::traits::{DeviceTrait, HostTrait, StreamTrait};
+use cpal::{
+    traits::{DeviceTrait, HostTrait, StreamTrait},
+    FromSample, SampleFormat, SizedSample,
+};
 use ringbuf::{
     storage::Heap,
     traits::{Consumer, Observer, Producer, Split},
@@ -178,24 +181,67 @@ fn build_input_stream(
     producer: AudioProducer,
     state_inner: Arc<VoiceStateInner>,
 ) -> Result<cpal::Stream, String> {
+    let sample_format = device
+        .default_input_config()
+        .map_err(|e| format!("Failed to get input config: {e}"))?
+        .sample_format();
+
+    match sample_format {
+        SampleFormat::F32 => build_input_stream_with::<f32>(device, config, producer, state_inner),
+        SampleFormat::I16 => build_input_stream_with::<i16>(device, config, producer, state_inner),
+        SampleFormat::I32 => build_input_stream_with::<i32>(device, config, producer, state_inner),
+        SampleFormat::I64 => build_input_stream_with::<i64>(device, config, producer, state_inner),
+        SampleFormat::U8 => build_input_stream_with::<u8>(device, config, producer, state_inner),
+        SampleFormat::U16 => build_input_stream_with::<u16>(device, config, producer, state_inner),
+        SampleFormat::U32 => build_input_stream_with::<u32>(device, config, producer, state_inner),
+        SampleFormat::U64 => build_input_stream_with::<u64>(device, config, producer, state_inner),
+        format => Err(format!("Unsupported input sample format: {format}")),
+    }
+}
+
+fn build_input_stream_with<T>(
+    device: &cpal::Device,
+    config: cpal::StreamConfig,
+    producer: AudioProducer,
+    state_inner: Arc<VoiceStateInner>,
+) -> Result<cpal::Stream, String>
+where
+    T: SizedSample + Send + 'static,
+    f32: FromSample<T>,
+{
     let native_sample_rate = config.sample_rate;
     let channels = config.channels as usize;
     let mut resampler = LinearResampler::new();
     let mut mono_buffer = Vec::with_capacity(2048);
     let mut resampled_buffer = Vec::with_capacity(2048);
+    let mut callback_count: u64 = 0;
 
     let inner_input_err = Arc::clone(&state_inner);
 
     device
         .build_input_stream(
             config,
-            move |data: &[f32], _| {
+            move |data: &[T], _| {
+                callback_count += 1;
+                if callback_count % 200 == 0 {
+                    let peak: f32 = data
+                        .iter()
+                        .map(|&s| s.to_sample::<f32>().abs())
+                        .fold(0.0f32, f32::max);
+                    eprintln!(
+                        "[vc] input callback #{callback_count}: {} frames, {} ch, peak {:.4}",
+                        data.len() / channels,
+                        channels,
+                        peak
+                    );
+                }
+
                 mono_buffer.clear();
                 resampled_buffer.clear();
 
-                // Downmix channels to mono
+                // Downmix channels to mono, converting to f32
                 for chunk in data.chunks_exact(channels) {
-                    let sum: f32 = chunk.iter().sum();
+                    let sum: f32 = chunk.iter().map(|&s| s.to_sample::<f32>()).sum();
                     mono_buffer.push(sum / channels as f32);
                 }
 
@@ -213,13 +259,21 @@ fn build_input_stream(
             },
             move |err| {
                 eprintln!("[vc] Input error: {err}. Attempting recovery...");
-                if let Ok(mut lock) = inner_input_err.session.lock() {
+                let rec = Arc::clone(&inner_input_err);
+                thread::spawn(move || {
+                    // Give the errored stream a moment to unwind before rebuilding.
+                    thread::sleep(Duration::from_millis(10));
+                    let Ok(mut lock) = rec.session.lock() else {
+                        return;
+                    };
                     if let Some(session) = lock.as_mut() {
+                        if session.shutdown.load(Ordering::SeqCst) {
+                            return;
+                        }
                         let target_device = session.current_input_device.clone();
-                        let _ = session
+                        let _ = session.update_input_device(target_device, Arc::clone(&rec));
-                            .update_input_device(target_device, Arc::clone(&inner_input_err));
                     }
-                }
+                });
             },
             None,
         )
@@ -233,25 +287,64 @@ fn build_output_stream(
     consumer: AudioConsumer,
     state_inner: Arc<VoiceStateInner>,
 ) -> Result<cpal::Stream, String> {
+    let sample_format = device
+        .default_output_config()
+        .map_err(|e| format!("Failed to get output config: {e}"))?
+        .sample_format();
+
+    match sample_format {
+        SampleFormat::F32 => build_output_stream_with::<f32>(device, config, consumer, state_inner),
+        SampleFormat::I16 => build_output_stream_with::<i16>(device, config, consumer, state_inner),
+        SampleFormat::I32 => build_output_stream_with::<i32>(device, config, consumer, state_inner),
+        SampleFormat::I64 => build_output_stream_with::<i64>(device, config, consumer, state_inner),
+        SampleFormat::U8 => build_output_stream_with::<u8>(device, config, consumer, state_inner),
+        SampleFormat::U16 => build_output_stream_with::<u16>(device, config, consumer, state_inner),
+        SampleFormat::U32 => build_output_stream_with::<u32>(device, config, consumer, state_inner),
+        SampleFormat::U64 => build_output_stream_with::<u64>(device, config, consumer, state_inner),
+        format => Err(format!("Unsupported output sample format: {format}")),
+    }
+}
+
+fn build_output_stream_with<T>(
+    device: &cpal::Device,
+    config: cpal::StreamConfig,
+    consumer: AudioConsumer,
+    state_inner: Arc<VoiceStateInner>,
+) -> Result<cpal::Stream, String>
+where
+    T: SizedSample + Send + 'static,
+    T: FromSample<f32>,
+{
     let native_sample_rate = config.sample_rate;
     let channels = config.channels as usize;
     let mut resampler = LinearResampler::new();
     let mut raw_mono_samples = Vec::with_capacity(2048);
     let mut resampled_mono = Vec::with_capacity(2048);
     let mut last_sample = 0.0f32;
+    let mut callback_count: u64 = 0;
 
     let inner_output_err = Arc::clone(&state_inner);
 
     device
         .build_output_stream(
             config,
-            move |data: &mut [f32], _| {
+            move |data: &mut [T], _| {
+                callback_count += 1;
+                if callback_count % 200 == 0 {
+                    eprintln!(
+                        "[vc] output callback #{callback_count}: {} frames, {} ch",
+                        data.len() / channels,
+                        channels
+                    );
+                }
+
                 let required_mono_samples = (data.len() / channels) * TARGET_SAMPLE_RATE as usize
                     / native_sample_rate as usize;
 
                 raw_mono_samples.clear();
                 resampled_mono.clear();
 
+                let mut underflow_count = 0usize;
                 if let Ok(mut cons) = consumer.lock() {
                     for _ in 0..required_mono_samples {
                         if let Some(s) = cons.try_pop() {
@@ -261,9 +354,18 @@ fn build_output_stream(
                             // Exponential decay to prevent clicking when underflowing
                             last_sample *= 0.92;
                             raw_mono_samples.push(last_sample);
+                            underflow_count += 1;
                         }
                     }
                 }
+                if callback_count % 200 == 0 {
+                    eprintln!(
+                        "[vc] output: got {}/{} mono samples ({} underflow)",
+                        required_mono_samples - underflow_count,
+                        required_mono_samples,
+                        underflow_count
+                    );
+                }
 
                 // Resample from 48kHz mono to target native output rate
                 resampler.process(
@@ -280,7 +382,7 @@ fn build_output_stream(
                     let mono_val = resampled_mono[res_idx];
                     for ch in 0..channels {
                         if out_idx + ch < data.len() {
-                            data[out_idx + ch] = mono_val;
+                            data[out_idx + ch] = T::from_sample(mono_val);
                         }
                     }
                     out_idx += channels;
@@ -289,13 +391,20 @@ fn build_output_stream(
             },
             move |err| {
                 eprintln!("[vc] Output error: {err}. Attempting recovery...");
-                if let Ok(mut lock) = inner_output_err.session.lock() {
+                let rec = Arc::clone(&inner_output_err);
+                thread::spawn(move || {
+                    thread::sleep(Duration::from_millis(10));
+                    let Ok(mut lock) = rec.session.lock() else {
+                        return;
+                    };
                     if let Some(session) = lock.as_mut() {
+                        if session.shutdown.load(Ordering::SeqCst) {
+                            return;
+                        }
                         let target_device = session.current_output_device.clone();
-                        let _ = session
+                        let _ = session.update_output_device(target_device, Arc::clone(&rec));
-                            .update_output_device(target_device, Arc::clone(&inner_output_err));
                     }
-                }
+                });
             },
             None,
         )
@@ -324,18 +433,22 @@ pub fn list_output_devices() -> Result<Vec<String>, String> {
 
 #[tauri::command]
 pub fn disconnect_from_vc(voice_state: State<'_, VoiceState>) -> Result<(), String> {
-    let mut lock = voice_state
+    let session = {
-        .inner
+        let mut lock = voice_state
-        .session
+            .inner
-        .lock()
+            .session
-        .map_err(|e| e.to_string())?;
+            .lock()
+            .map_err(|e| e.to_string())?;
+        lock.take()
+    };
 
-    if let Some(session) = lock.take() {
+    if let Some(session) = session {
         session.shutdown.store(true, Ordering::SeqCst);
         let _ = session.input_stream.pause();
         if let Ok(output) = session.output_stream.lock() {
             let _ = output.pause();
         }
+        eprintln!("[vc] disconnected and paused streams");
     }
 
     Ok(())
@@ -349,59 +462,102 @@ pub fn connect_to_vc(
     config_state: State<'_, ConfigState>,
     voice_state: State<'_, VoiceState>,
 ) -> Result<(), String> {
-    disconnect_from_vc(voice_state.clone())?;
+    eprintln!("[vc] connect_to_vc called hostname={hostname} pin={pin}");
 
-    let key = Key::try_from(shared_secret.as_slice()).map_err(|v| v.to_string())?;
+    if let Err(e) = disconnect_from_vc(voice_state.clone()) {
+        eprintln!("[vc] disconnect_from_vc failed: {e}");
+    }
+
+    let key = match Key::try_from(shared_secret.as_slice()) {
+        Ok(k) => k,
+        Err(v) => {
+            let msg = v.to_string();
+            eprintln!("[vc] invalid key length: {msg}");
+            return Err(msg);
+        }
+    };
     let cipher = Arc::new(Mutex::new(SessionCipher::new(ChaCha20Poly1305::new(&key))));
 
     let config = config_state.0.lock().unwrap().clone();
     let state_inner = Arc::clone(&voice_state.inner);
 
-    let socket = Arc::new(
+    let socket = match UdpSocket::bind("0.0.0.0:0") {
-        UdpSocket::bind("0.0.0.0:0").map_err(|e| format!("Failed to bind UDP socket: {e}"))?,
+        Ok(s) => s,
-    );
+        Err(e) => {
-    socket
+            let msg = format!("Failed to bind UDP socket: {e}");
-        .connect(&hostname)
+            eprintln!("[vc] {msg}");
-        .map_err(|e| format!("Failed to connect UDP socket: {e}"))?;
+            return Err(msg);
-    socket
+        }
-        .set_read_timeout(Some(Duration::from_millis(5)))
+    };
-        .map_err(|e| e.to_string())?;
+    let socket = Arc::new(socket);
-    socket
+    if let Err(e) = socket.connect(&hostname) {
-        .send(&pin.to_be_bytes())
+        let msg = format!("Failed to connect UDP socket: {e}");
-        .map_err(|e| format!("Failed to send pin: {e}"))?;
+        eprintln!("[vc] {msg}");
+        return Err(msg);
+    }
+    if let Err(e) = socket.set_read_timeout(Some(Duration::from_millis(5))) {
+        let msg = e.to_string();
+        eprintln!("[vc] set_read_timeout failed: {msg}");
+        return Err(msg);
+    }
+    if let Err(e) = socket.send(&pin.to_be_bytes()) {
+        let msg = format!("Failed to send pin: {e}");
+        eprintln!("[vc] {msg}");
+        return Err(msg);
+    }
 
     let host = cpal::default_host();
 
     let input_device = match &config.input_device_name {
         Some(name) => host
             .input_devices()
-            .map_err(|e| e.to_string())?
+            .map_err(|e| {
+                let msg = e.to_string();
+                eprintln!("[vc] failed to enumerate input devices: {msg}");
+                msg
+            })?
             .find(|d| {
                 d.description()
                     .ok()
                     .map(|x| x.name() == *name)
                     .unwrap_or(false)
             })
-            .ok_or("Input device not found")?,
+            .ok_or_else(|| {
-        None => host
+                let msg = format!("Input device not found: {name}");
-            .default_input_device()
+                eprintln!("[vc] {msg}");
-            .ok_or("No default input device")?,
+                msg
+            })?,
+        None => host.default_input_device().ok_or_else(|| {
+            let msg = "No default input device".to_string();
+            eprintln!("[vc] {msg}");
+            msg
+        })?,
     };
 
     let output_device = match &config.output_device_name {
         Some(name) => host
             .output_devices()
-            .map_err(|e| e.to_string())?
+            .map_err(|e| {
+                let msg = e.to_string();
+                eprintln!("[vc] failed to enumerate output devices: {msg}");
+                msg
+            })?
             .find(|d| {
                 d.description()
                     .ok()
                     .map(|x| x.name() == *name)
                     .unwrap_or(false)
             })
-            .ok_or("Output device not found")?,
+            .ok_or_else(|| {
-        None => host
+                let msg = format!("Output device not found: {name}");
-            .default_output_device()
+                eprintln!("[vc] {msg}");
-            .ok_or("No default output device")?,
+                msg
+            })?,
+        None => host.default_output_device().ok_or_else(|| {
+            let msg = "No default output device".to_string();
+            eprintln!("[vc] {msg}");
+            msg
+        })?,
     };
 
     let shutdown = Arc::new(AtomicBool::new(false));
@@ -417,6 +573,12 @@ pub fn connect_to_vc(
         .map_err(|e| format!("Failed to get default output config: {e}"))?
         .config();
 
+    eprintln!(
+        "[vc] Output device {:?} config: {} Hz, {} ch",
+        output_device.description().map(|d| d.name().to_string()),
+        output_config.sample_rate,
+        output_config.channels
+    );
     let output_stream = build_output_stream(
         &output_device,
         output_config,
@@ -437,6 +599,12 @@ pub fn connect_to_vc(
         .map_err(|e| format!("Failed to get default input config: {e}"))?
         .config();
 
+    eprintln!(
+        "[vc] Input device {:?} config: {} Hz, {} ch",
+        input_device.description().map(|d| d.name().to_string()),
+        input_config.sample_rate,
+        input_config.channels
+    );
     let input_stream = build_input_stream(
         &input_device,
         input_config,
@@ -454,6 +622,9 @@ pub fn connect_to_vc(
             let mut sequence = 0u32;
             let mut frame_buf = vec![0.0f32; PACKET_SAMPLES];
             let mut hangover_counter = 0;
+            let mut loop_count: u64 = 0;
+            // Adaptive VAD: track a slow-moving noise floor so quiet mics still trigger.
+            let mut noise_floor: f32 = 0.0;
 
             while !shutdown.load(Ordering::Relaxed) {
                 if consumer_in.occupied_len() >= PACKET_SAMPLES {
@@ -462,7 +633,17 @@ pub fn connect_to_vc(
                     let sum_squares: f32 = frame_buf.iter().map(|&s| s * s).sum();
                     let rms = (sum_squares / PACKET_SAMPLES as f32).sqrt();
 
-                    let is_speaking = if rms >= VAD_THRESHOLD {
+                    // Update noise floor (attack fast, release slow).
+                    if noise_floor == 0.0 {
+                        noise_floor = rms;
+                    } else if rms < noise_floor {
+                        noise_floor = noise_floor * 0.9 + rms * 0.1;
+                    } else {
+                        noise_floor = noise_floor * 0.999;
+                    }
+                    let threshold = (noise_floor * 4.0).max(VAD_THRESHOLD);
+
+                    let is_speaking = if rms >= threshold {
                         hangover_counter = VAD_HANGOVER_FRAMES;
                         true
                     } else if hangover_counter > 0 {
@@ -472,6 +653,14 @@ pub fn connect_to_vc(
                         false
                     };
 
+                    loop_count += 1;
+                    if loop_count % 200 == 0 {
+                        eprintln!(
+                            "[vc] sender: buffered {} samples, frame rms {rms:.4}, floor {noise_floor:.4}, thr {threshold:.4}, speaking {is_speaking}, seq {sequence}",
+                            consumer_in.occupied_len()
+                        );
+                    }
+
                     if is_speaking {
                         // sequence goes INSIDE the plaintext now, prefixed before the PCM
                         let mut plaintext = Vec::with_capacity(4 + PACKET_SAMPLES * 2);
@@ -488,7 +677,15 @@ pub fn connect_to_vc(
                             continue;
                         };
 
-                        let _ = input_socket.send(&net_packet);
+                        let sent = input_socket.send(&net_packet);
+                        if let Err(e) = sent {
+                            eprintln!("[vc] sender: send failed: {e}");
+                        } else if loop_count % 200 == 0 {
+                            eprintln!(
+                                "[vc] sender: sent packet seq {sequence} ({} bytes)",
+                                net_packet.len()
+                            );
+                        }
                         sequence = sequence.wrapping_add(1);
                     }
                 } else {
@@ -511,9 +708,13 @@ pub fn connect_to_vc(
             let mut last_good_frame = vec![0.0f32; PACKET_SAMPLES];
             let mut udp_buffer = [0u8; MAX_PACKET_SIZE];
             let mut next_frame_time = Instant::now();
+            let mut recv_count: u64 = 0;
+            let mut loop_count: u64 = 0;
 
             while !shutdown.load(Ordering::Relaxed) {
+                loop_count += 1;
                 if let Ok(len) = socket.recv(&mut udp_buffer) {
+                    recv_count += 1;
                     match cipher.lock().unwrap().decrypt(&udp_buffer[..len]) {
                         Ok(plaintext) => {
                             if plaintext.len() < 4 {
@@ -536,6 +737,13 @@ pub fn connect_to_vc(
                     }
                 }
 
+                if loop_count % 2000 == 0 {
+                    eprintln!(
+                        "[vc] receiver: received {recv_count} packets total, {} buffered, prebuffering {is_prebuffering}, expected {expected:?}",
+                        packets.len()
+                    );
+                }
+
                 if is_prebuffering {
                     if packets.len() >= INITIAL_PACKET_CUSHION {
                         expected = packets.keys().next().copied();

src-tauri/src/crypto.rs

@@ -3,7 +3,6 @@ use chacha20poly1305::{aead::Aead, ChaCha20Poly1305, Nonce};
 pub struct SessionCipher {
     cipher: ChaCha20Poly1305,
     send_counter: u64,
-    recv_counter: u64,
 }
 
 impl SessionCipher {
@@ -11,11 +10,10 @@ impl SessionCipher {
         Self {
             cipher,
             send_counter: 0,
-            recv_counter: 0,
         }
     }
 
-    pub fn next_send_nonce(&mut self) -> [u8; 12] {
+    fn next_send_nonce(&mut self) -> [u8; 12] {
         let mut nonce = [0u8; 12];
         nonce[..8].copy_from_slice(&self.send_counter.to_be_bytes());
         // top bit distinguishes "send" direction from "recv" direction,
@@ -26,13 +24,6 @@ impl SessionCipher {
         nonce
     }
 
-    pub fn next_recv_nonce(&mut self) -> [u8; 12] {
-        let mut nonce = [0u8; 12];
-        nonce[..8].copy_from_slice(&self.recv_counter.to_be_bytes());
-        self.recv_counter += 1;
-        nonce
-    }
-
     pub fn encrypt(&mut self, plaintext: &[u8]) -> Result<Vec<u8>, String> {
         let nonce_bytes = self.next_send_nonce();
         let nonce = Nonce::try_from(nonce_bytes).map_err(|v| v.to_string())?;
@@ -48,11 +39,15 @@ impl SessionCipher {
         Ok(out)
     }
 
-    pub fn decrypt(&mut self, data: &[u8]) -> Result<Vec<u8>, String> {
+    pub fn decrypt(&self, data: &[u8]) -> Result<Vec<u8>, String> {
         if data.len() < 12 {
             return Err("message too short to contain a nonce".to_string());
         }
         let (nonce_bytes, ciphertext) = data.split_at(12);
+
+        // Use the nonce that travelled with this packet rather than a locally
+        // tracked counter. This makes decryption immune to UDP packet loss or
+        // reordering, because we never desync from the sender's counter.
         let nonce = Nonce::try_from(nonce_bytes).map_err(|v| v.to_string())?;
 
         self.cipher