Revert "add jitter buffer for voice"

This reverts commit 3306edc.
uowuo · Apr 4, 2024 · 72b6c3a · 72b6c3a
1 parent fda687e
commit 72b6c3a
Show file tree

Hide file tree

Showing 6 changed files with 11 additions and 110 deletions.
diff --git a/README.md b/README.md
@@ -330,12 +330,10 @@ For example, memory_db would be set by adding `memory_db = true` under the line
 
 #### voice
 
-| Setting                  | Type   | Default                            | Description                                                                                                                |
-|--------------------------|--------|------------------------------------|----------------------------------------------------------------------------------------------------------------------------|
-| `vad`                    | string | rnnoise if enabled, gate otherwise | Method used for voice activity detection. Changeable in UI                                                                 |
-| `backends`               | string | empty                              | Change backend priority when initializing miniaudio: `wasapi;dsound;winmm;coreaudio;sndio;audio4;oss;pulseaudio;alsa;jack` |
-| `jitter_latency_desired` | int    | 50                                 | Desired/Minimum latency for jitter buffer (in milliseconds)                                                                |
-| `jitter_latency_maximum` | int    | 200                                | Maximum latency for jitter buffer before frames are discarded (in milliseconds)                                            |
+| Setting    | Type   | Default                            | Description                                                                                                                |
+|------------|--------|------------------------------------|----------------------------------------------------------------------------------------------------------------------------|
+| `vad`      | string | rnnoise if enabled, gate otherwise | Method used for voice activity detection. Changeable in UI                                                                 |
+| `backends` | string | empty                              | Change backend priority when initializing miniaudio: `wasapi;dsound;winmm;coreaudio;sndio;audio4;oss;pulseaudio;alsa;jack` |
 
 #### windows
 

diff --git a/src/audio/jitterbuffer.hpp b/src/audio/jitterbuffer.hpp
diff --git a/src/audio/manager.cpp b/src/audio/manager.cpp
@@ -26,7 +26,6 @@ const uint8_t *StripRTPExtensionHeader(const uint8_t *buf, int num_bytes, size_t
     return buf;
 }
 
-// frameCount is configured to be 480 samples per channel
 void data_callback(ma_device *pDevice, void *pOutput, const void *pInput, ma_uint32 frameCount) {
     AudioManager *mgr = reinterpret_cast<AudioManager *>(pDevice->pUserData);
     if (mgr == nullptr) return;
@@ -38,14 +37,12 @@ void data_callback(ma_device *pDevice, void *pOutput, const void *pInput, ma_uin
         if (const auto vol_it = mgr->m_volume_ssrc.find(ssrc); vol_it != mgr->m_volume_ssrc.end()) {
             volume = vol_it->second;
         }
-
-        static std::array<int16_t, 480 * 2> buf;
-
-        if (!pair.first.PopSamples(buf.data(), 480 * 2)) continue;
-
-        for (size_t i = 0; i < 480 * 2; i++) {
+        auto &buf = pair.first;
+        const size_t n = std::min(static_cast<size_t>(buf.size()), static_cast<size_t>(frameCount * 2ULL));
+        for (size_t i = 0; i < n; i++) {
             pOutputF32[i] += volume * buf[i] / 32768.F;
         }
+        buf.erase(buf.begin(), buf.begin() + n);
     }
 }
 
@@ -205,14 +202,7 @@ void AudioManager::AddSSRC(uint32_t ssrc) {
     int error;
     if (m_sources.find(ssrc) == m_sources.end()) {
         auto *decoder = opus_decoder_create(48000, 2, &error);
-        auto &s = Abaddon::Get().GetSettings();
-        m_sources.insert(std::make_pair(ssrc, std::make_pair(
-                                                  JitterBuffer<int16_t>(
-                                                      s.JitterDesiredLatency,
-                                                      s.JitterMaximumLatency,
-                                                      2,
-                                                      48000),
-                                                  decoder)));
+        m_sources.insert(std::make_pair(ssrc, std::make_pair(std::deque<int16_t> {}, decoder)));
     }
 }
 
@@ -252,7 +242,7 @@ void AudioManager::FeedMeOpus(uint32_t ssrc, const std::vector<uint8_t> &data) {
         } else {
             UpdateReceiveVolume(ssrc, pcm.data(), decoded);
             auto &buf = it->second.first;
-            buf.PushSamples(pcm.data(), decoded * 2);
+            buf.insert(buf.end(), pcm.begin(), pcm.begin() + decoded * 2);
         }
     }
 }

diff --git a/src/audio/manager.hpp b/src/audio/manager.hpp
@@ -21,7 +21,6 @@
 #endif
 
 #include "devices.hpp"
-#include "jitterbuffer.hpp"
 // clang-format on
 
 class AudioManager {
@@ -137,7 +136,7 @@ class AudioManager {
     mutable std::mutex m_rnn_mutex;
 #endif
 
-    std::unordered_map<uint32_t, std::pair<JitterBuffer<int16_t>, OpusDecoder *>> m_sources;
+    std::unordered_map<uint32_t, std::pair<std::deque<int16_t>, OpusDecoder *>> m_sources;
 
     OpusEncoder *m_encoder;
 

diff --git a/src/settings.cpp b/src/settings.cpp
@@ -130,8 +130,6 @@ void SettingsManager::DefineSettings() {
     AddSetting("voice", "vad", "gate"s, &Settings::VAD);
 #endif
     AddSetting("voice", "backends", ""s, &Settings::Backends);
-    AddSetting("voice", "jitter_latency_desired", 50, &Settings::JitterDesiredLatency);
-    AddSetting("voice", "jitter_latency_maximum", 200, &Settings::JitterMaximumLatency);
 }
 
 void SettingsManager::ReadSettings() {

diff --git a/src/settings.hpp b/src/settings.hpp
@@ -52,8 +52,6 @@ class SettingsManager {
         // [voice]
         std::string VAD;
         std::string Backends;
-        int JitterDesiredLatency;
-        int JitterMaximumLatency;
 
         // [windows]
         bool HideConsole;