Increase default audio buffer for FLAC

FLAC is not low latency: the default encoder produces blocks of 4096
samples, which represent ~85.333ms.

Increase the audio buffer by default so that audio playback works.

app/data/bash-completion/scrcpy

@@ -125,7 +125,7 @@ _scrcpy() {
             return
             ;;
         --record-format)
-            COMPREPLY=($(compgen -W 'mp4 mkv m4a mka opus aac flac wav' -- "$cur"))
+            COMPREPLY=($(compgen -W 'mkv mp4' -- "$cur"))
             return
             ;;
         --render-driver)

app/data/zsh-completion/_scrcpy

@@ -65,7 +65,7 @@ arguments=(
     '--push-target=[Set the target directory for pushing files to the device by drag and drop]'
     {-r,--record=}'[Record screen to file]:record file:_files'
     '--raw-key-events[Inject key events for all input keys, and ignore text events]'
-    '--record-format=[Force recording format]:format:(mp4 mkv m4a mka opus aac flac wav)'
+    '--record-format=[Force recording format]:format:(mp4 mkv)'
     '--render-driver=[Request SDL to use the given render driver]:driver name:(direct3d opengl opengles2 opengles metal software)'
     '--require-audio=[Make scrcpy fail if audio is enabled but does not work]'
     '--rotation=[Set the initial display rotation]:rotation values:(0 1 2 3)'

app/prebuilt-deps/prepare-ffmpeg.sh

@@ -6,11 +6,11 @@ cd "$DIR"
 mkdir -p "$PREBUILT_DATA_DIR"
 cd "$PREBUILT_DATA_DIR"
 
-VERSION=6.1-scrcpy-2
+VERSION=6.1-scrcpy
 DEP_DIR="ffmpeg-$VERSION"
 
 FILENAME="$DEP_DIR".7z
-SHA256SUM=7f25f638dc24a0f5d4af07a088b6a604cf33548900bbfd2f6ce0bae050b7664d
+SHA256SUM=b41726e603f4624bb9ed7d2836e3e59d9d20b000e22a9ebd27055f4e99e48219
 
 if [[ -d "$DEP_DIR" ]]
 then

app/scrcpy.1

@@ -347,7 +347,7 @@ Record screen to
 
 The format is determined by the
 .B \-\-record\-format
-option if set, or by the file extension.
+option if set, or by the file extension (.mp4 or .mkv).
 
 .TP
 .B \-\-raw\-key\-events
@@ -355,7 +355,7 @@ Inject key events for all input keys, and ignore text events.
 
 .TP
 .BI "\-\-record\-format " format
-Force recording format (mp4, mkv, m4a, mka, opus, aac, flac or wav).
+Force recording format (either mp4 or mkv).
 
 .TP
 .BI "\-\-render\-driver " name

app/src/cli.c

@@ -583,7 +583,7 @@ static const struct sc_option options[] = {
         .argdesc = "file.mp4",
         .text = "Record screen to file.\n"
                 "The format is determined by the --record-format option if "
-                "set, or by the file extension.",
+                "set, or by the file extension (.mp4 or .mkv).",
     },
     {
         .longopt_id = OPT_RAW_KEY_EVENTS,
@@ -594,8 +594,7 @@ static const struct sc_option options[] = {
         .longopt_id = OPT_RECORD_FORMAT,
         .longopt = "record-format",
         .argdesc = "format",
-        .text = "Force recording format (mp4, mkv, m4a, mka, opus, aac, flac "
-                "or wav).",
+        .text = "Force recording format (either mp4 or mkv).",
     },
     {
         .longopt_id = OPT_RENDER_DRIVER,
@@ -1630,9 +1629,6 @@ get_record_format(const char *name) {
     if (!strcmp(name, "flac")) {
         return SC_RECORD_FORMAT_FLAC;
     }
-    if (!strcmp(name, "wav")) {
-        return SC_RECORD_FORMAT_WAV;
-    }
     return 0;
 }
 
@@ -2376,6 +2372,11 @@ parse_args_with_getopt(struct scrcpy_cli_args *args, int argc, char *argv[],
             }
         }
 
+        if (opts->audio_codec == SC_CODEC_RAW) {
+            LOGE("Recording does not support RAW audio codec");
+            return false;
+        }
+
         if (opts->video
                 && sc_record_format_is_audio_only(opts->record_format)) {
             LOGE("Audio container does not support video stream");
@@ -2401,20 +2402,6 @@ parse_args_with_getopt(struct scrcpy_cli_args *args, int argc, char *argv[],
                  "(try with --audio-codec=flac)");
             return false;
         }
-
-        if (opts->record_format == SC_RECORD_FORMAT_WAV
-                && opts->audio_codec != SC_CODEC_RAW) {
-            LOGE("Recording to WAV file requires a RAW audio stream "
-                 "(try with --audio-codec=raw)");
-            return false;
-        }
-
-        if ((opts->record_format == SC_RECORD_FORMAT_MP4 ||
-             opts->record_format == SC_RECORD_FORMAT_M4A)
-                && opts->audio_codec == SC_CODEC_RAW) {
-            LOGE("Recording to MP4 container does not support RAW audio");
-            return false;
-        }
     }
 
     if (opts->audio_codec == SC_CODEC_FLAC && opts->audio_bit_rate) {

app/src/options.h

@@ -26,7 +26,6 @@ enum sc_record_format {
     SC_RECORD_FORMAT_OPUS,
     SC_RECORD_FORMAT_AAC,
     SC_RECORD_FORMAT_FLAC,
-    SC_RECORD_FORMAT_WAV,
 };
 
 static inline bool
@@ -35,8 +34,7 @@ sc_record_format_is_audio_only(enum sc_record_format fmt) {
         || fmt == SC_RECORD_FORMAT_MKA
         || fmt == SC_RECORD_FORMAT_OPUS
         || fmt == SC_RECORD_FORMAT_AAC
-        || fmt == SC_RECORD_FORMAT_FLAC
-        || fmt == SC_RECORD_FORMAT_WAV;
+        || fmt == SC_RECORD_FORMAT_FLAC;
 }
 
 enum sc_codec {

app/src/recorder.c

@@ -71,8 +71,6 @@ sc_recorder_get_format_name(enum sc_record_format format) {
             return "opus";
         case SC_RECORD_FORMAT_FLAC:
             return "flac";
-        case SC_RECORD_FORMAT_WAV:
-            return "wav";
         default:
             return NULL;
     }
@@ -105,7 +103,7 @@ sc_recorder_write_stream(struct sc_recorder *recorder,
     AVStream *stream = recorder->ctx->streams[st->index];
     sc_recorder_rescale_packet(stream, packet);
     if (st->last_pts != AV_NOPTS_VALUE && packet->pts <= st->last_pts) {
-        LOGD("Fixing PTS non monotonically increasing in stream %d "
+        LOGW("Fixing PTS non monotonically increasing in stream %d "
              "(%" PRIi64 " >= %" PRIi64 ")",
              st->index, st->last_pts, packet->pts);
         packet->pts = ++st->last_pts;
@@ -170,14 +168,13 @@ sc_recorder_close_output_file(struct sc_recorder *recorder) {
 }
 
 static inline bool
-sc_recorder_must_wait_for_config_packets(struct sc_recorder *recorder) {
+sc_recorder_has_empty_queues(struct sc_recorder *recorder) {
     if (recorder->video && sc_vecdeque_is_empty(&recorder->video_queue)) {
         // The video queue is empty
         return true;
     }
 
-    if (recorder->audio && recorder->audio_expects_config_packet
-            && sc_vecdeque_is_empty(&recorder->audio_queue)) {
+    if (recorder->audio && sc_vecdeque_is_empty(&recorder->audio_queue)) {
         // The audio queue is empty (when audio is enabled)
         return true;
     }
@@ -193,7 +190,7 @@ sc_recorder_process_header(struct sc_recorder *recorder) {
     while (!recorder->stopped &&
               ((recorder->video && !recorder->video_init)
             || (recorder->audio && !recorder->audio_init)
-            || sc_recorder_must_wait_for_config_packets(recorder))) {
+            || sc_recorder_has_empty_queues(recorder))) {
         sc_cond_wait(&recorder->cond, &recorder->mutex);
     }
 
@@ -212,8 +209,7 @@ sc_recorder_process_header(struct sc_recorder *recorder) {
     }
 
     AVPacket *audio_pkt = NULL;
-    if (recorder->audio_expects_config_packet &&
-            !sc_vecdeque_is_empty(&recorder->audio_queue)) {
+    if (!sc_vecdeque_is_empty(&recorder->audio_queue)) {
         assert(recorder->audio);
         audio_pkt = sc_vecdeque_pop(&recorder->audio_queue);
     }
@@ -601,10 +597,6 @@ sc_recorder_audio_packet_sink_open(struct sc_packet_sink *sink,
 
     recorder->audio_stream.index = stream->index;
 
-    // A config packet is provided for all formats supported except raw audio
-    recorder->audio_expects_config_packet =
-        ctx->codec_id != AV_CODEC_ID_PCM_S16LE;
-
     recorder->audio_init = true;
     sc_cond_signal(&recorder->cond);
     sc_mutex_unlock(&recorder->mutex);
@@ -717,8 +709,6 @@ sc_recorder_init(struct sc_recorder *recorder, const char *filename,
     recorder->video_init = false;
     recorder->audio_init = false;
 
-    recorder->audio_expects_config_packet = false;
-
     sc_recorder_stream_init(&recorder->video_stream);
     sc_recorder_stream_init(&recorder->audio_stream);
 

app/src/recorder.h

@@ -50,8 +50,6 @@ struct sc_recorder {
     bool video_init;
     bool audio_init;
 
-    bool audio_expects_config_packet;
-
     struct sc_recorder_stream video_stream;
     struct sc_recorder_stream audio_stream;
 

cross_win32.txt

@@ -16,6 +16,6 @@ cpu = 'i686'
 endian = 'little'
 
 [properties]
-prebuilt_ffmpeg = 'ffmpeg-6.1-scrcpy-2/win32'
+prebuilt_ffmpeg = 'ffmpeg-6.1-scrcpy/win32'
 prebuilt_sdl2 = 'SDL2-2.28.4/i686-w64-mingw32'
 prebuilt_libusb = 'libusb-1.0.26/libusb-MinGW-Win32'

cross_win64.txt

@@ -16,6 +16,6 @@ cpu = 'x86_64'
 endian = 'little'
 
 [properties]
-prebuilt_ffmpeg = 'ffmpeg-6.1-scrcpy-2/win64'
+prebuilt_ffmpeg = 'ffmpeg-6.1-scrcpy/win64'
 prebuilt_sdl2 = 'SDL2-2.28.4/x86_64-w64-mingw32'
 prebuilt_libusb = 'libusb-1.0.26/libusb-MinGW-x64'

doc/recording.md

@@ -19,7 +19,6 @@ To record only the audio:
 scrcpy --no-video --record=file.opus
 scrcpy --no-video --audio-codec=aac --record=file.aac
 scrcpy --no-video --audio-codec=flac --record=file.flac
-scrcpy --no-video --audio-codec=raw --record=file.wav
 # .m4a/.mp4 and .mka/.mkv are also supported for opus, aac and flac
 ```
 
@@ -33,17 +32,14 @@ course, not if you capture your scrcpy window and audio output on the computer).
 ## Format
 
 The video and audio streams are encoded on the device, but are muxed on the
-client side. Several formats (containers) are supported:
- - MP4 (`.mp4`, `.m4a`, `.aac`)
- - Matroska (`.mkv`, `.mka`)
- - OPUS (`.opus`)
- - FLAC (`.flac`)
- - WAV (`.wav`)
+client side. Two formats (containers) are supported:
+ - Matroska (`.mkv`)
+ - MP4 (`.mp4`)
 
 The container is automatically selected based on the filename.
 
 It is also possible to explicitly select a container (in that case the filename
-needs not end with a known extension):
+needs not end with `.mkv` or `.mp4`):
 
 ```
 scrcpy --record=file --record-format=mkv

release.mk

@@ -94,10 +94,10 @@ dist-win32: build-server build-win32
 	cp app/data/scrcpy-noconsole.vbs "$(DIST)/$(WIN32_TARGET_DIR)"
 	cp app/data/icon.png "$(DIST)/$(WIN32_TARGET_DIR)"
 	cp app/data/open_a_terminal_here.bat "$(DIST)/$(WIN32_TARGET_DIR)"
-	cp app/prebuilt-deps/data/ffmpeg-6.1-scrcpy-2/win32/bin/avutil-58.dll "$(DIST)/$(WIN32_TARGET_DIR)/"
-	cp app/prebuilt-deps/data/ffmpeg-6.1-scrcpy-2/win32/bin/avcodec-60.dll "$(DIST)/$(WIN32_TARGET_DIR)/"
-	cp app/prebuilt-deps/data/ffmpeg-6.1-scrcpy-2/win32/bin/avformat-60.dll "$(DIST)/$(WIN32_TARGET_DIR)/"
-	cp app/prebuilt-deps/data/ffmpeg-6.1-scrcpy-2/win32/bin/swresample-4.dll "$(DIST)/$(WIN32_TARGET_DIR)/"
+	cp app/prebuilt-deps/data/ffmpeg-6.1-scrcpy/win32/bin/avutil-58.dll "$(DIST)/$(WIN32_TARGET_DIR)/"
+	cp app/prebuilt-deps/data/ffmpeg-6.1-scrcpy/win32/bin/avcodec-60.dll "$(DIST)/$(WIN32_TARGET_DIR)/"
+	cp app/prebuilt-deps/data/ffmpeg-6.1-scrcpy/win32/bin/avformat-60.dll "$(DIST)/$(WIN32_TARGET_DIR)/"
+	cp app/prebuilt-deps/data/ffmpeg-6.1-scrcpy/win32/bin/swresample-4.dll "$(DIST)/$(WIN32_TARGET_DIR)/"
 	cp app/prebuilt-deps/data/platform-tools-34.0.5/adb.exe "$(DIST)/$(WIN32_TARGET_DIR)/"
 	cp app/prebuilt-deps/data/platform-tools-34.0.5/AdbWinApi.dll "$(DIST)/$(WIN32_TARGET_DIR)/"
 	cp app/prebuilt-deps/data/platform-tools-34.0.5/AdbWinUsbApi.dll "$(DIST)/$(WIN32_TARGET_DIR)/"
@@ -112,10 +112,10 @@ dist-win64: build-server build-win64
 	cp app/data/scrcpy-noconsole.vbs "$(DIST)/$(WIN64_TARGET_DIR)"
 	cp app/data/icon.png "$(DIST)/$(WIN64_TARGET_DIR)"
 	cp app/data/open_a_terminal_here.bat "$(DIST)/$(WIN64_TARGET_DIR)"
-	cp app/prebuilt-deps/data/ffmpeg-6.1-scrcpy-2/win64/bin/avutil-58.dll "$(DIST)/$(WIN64_TARGET_DIR)/"
-	cp app/prebuilt-deps/data/ffmpeg-6.1-scrcpy-2/win64/bin/avcodec-60.dll "$(DIST)/$(WIN64_TARGET_DIR)/"
-	cp app/prebuilt-deps/data/ffmpeg-6.1-scrcpy-2/win64/bin/avformat-60.dll "$(DIST)/$(WIN64_TARGET_DIR)/"
-	cp app/prebuilt-deps/data/ffmpeg-6.1-scrcpy-2/win64/bin/swresample-4.dll "$(DIST)/$(WIN64_TARGET_DIR)/"
+	cp app/prebuilt-deps/data/ffmpeg-6.1-scrcpy/win64/bin/avutil-58.dll "$(DIST)/$(WIN64_TARGET_DIR)/"
+	cp app/prebuilt-deps/data/ffmpeg-6.1-scrcpy/win64/bin/avcodec-60.dll "$(DIST)/$(WIN64_TARGET_DIR)/"
+	cp app/prebuilt-deps/data/ffmpeg-6.1-scrcpy/win64/bin/avformat-60.dll "$(DIST)/$(WIN64_TARGET_DIR)/"
+	cp app/prebuilt-deps/data/ffmpeg-6.1-scrcpy/win64/bin/swresample-4.dll "$(DIST)/$(WIN64_TARGET_DIR)/"
 	cp app/prebuilt-deps/data/platform-tools-34.0.5/adb.exe "$(DIST)/$(WIN64_TARGET_DIR)/"
 	cp app/prebuilt-deps/data/platform-tools-34.0.5/AdbWinApi.dll "$(DIST)/$(WIN64_TARGET_DIR)/"
 	cp app/prebuilt-deps/data/platform-tools-34.0.5/AdbWinUsbApi.dll "$(DIST)/$(WIN64_TARGET_DIR)/"

server/src/main/java/com/genymobile/scrcpy/AudioCapture.java

@@ -24,19 +24,11 @@ public final class AudioCapture {
     public static final int ENCODING = AudioFormat.ENCODING_PCM_16BIT;
     public static final int BYTES_PER_SAMPLE = 2;
 
-    // Never read more than 1024 samples, even if the buffer is bigger (that would increase latency).
-    // A lower value is useless, since the system captures audio samples by blocks of 1024 (so for example if we read by blocks of 256 samples, we
-    // receive 4 successive blocks without waiting, then we wait for the 4 next ones).
-    public static final int MAX_READ_SIZE = 1024 * CHANNELS * BYTES_PER_SAMPLE;
-
-    private static final long ONE_SAMPLE_US = (1000000 + SAMPLE_RATE - 1) / SAMPLE_RATE; // 1 sample in microseconds (used for fixing PTS)
-
     private final int audioSource;
 
     private AudioRecord recorder;
 
     private final AudioTimestamp timestamp = new AudioTimestamp();
-    private long previousRecorderTimestamp = -1;
     private long previousPts = 0;
     private long nextPts = 0;
 
@@ -44,6 +36,10 @@ public final class AudioCapture {
         this.audioSource = audioSource.value();
     }
 
+    public static int millisToBytes(int millis) {
+        return SAMPLE_RATE * CHANNELS * BYTES_PER_SAMPLE * millis / 1000;
+    }
+
     private static AudioFormat createAudioFormat() {
         AudioFormat.Builder builder = new AudioFormat.Builder();
         builder.setEncoding(ENCODING);
@@ -139,8 +135,8 @@ public final class AudioCapture {
     }
 
     @TargetApi(Build.VERSION_CODES.N)
-    public int read(ByteBuffer directBuffer, MediaCodec.BufferInfo outBufferInfo) {
-        int r = recorder.read(directBuffer, MAX_READ_SIZE);
+    public int read(ByteBuffer directBuffer, int size, MediaCodec.BufferInfo outBufferInfo) {
+        int r = recorder.read(directBuffer, size);
         if (r <= 0) {
             return r;
         }
@@ -148,9 +144,8 @@ public final class AudioCapture {
         long pts;
 
         int ret = recorder.getTimestamp(timestamp, AudioTimestamp.TIMEBASE_MONOTONIC);
-        if (ret == AudioRecord.SUCCESS && timestamp.nanoTime != previousRecorderTimestamp) {
+        if (ret == AudioRecord.SUCCESS) {
             pts = timestamp.nanoTime / 1000;
-            previousRecorderTimestamp = timestamp.nanoTime;
         } else {
             if (nextPts == 0) {
                 Ln.w("Could not get any audio timestamp");
@@ -162,13 +157,13 @@ public final class AudioCapture {
         long durationUs = r * 1000000 / (CHANNELS * BYTES_PER_SAMPLE * SAMPLE_RATE);
         nextPts = pts + durationUs;
 
-        if (previousPts != 0 && pts < previousPts + ONE_SAMPLE_US) {
+        if (previousPts != 0 && pts < previousPts) {
             // Audio PTS may come from two sources:
             //  - recorder.getTimestamp() if the call works;
             //  - an estimation from the previous PTS and the packet size as a fallback.
             //
             // Therefore, the property that PTS are monotonically increasing is no guaranteed in corner cases, so enforce it.
-            pts = previousPts + ONE_SAMPLE_US;
+            pts = previousPts + 1;
         }
         previousPts = pts;
 

server/src/main/java/com/genymobile/scrcpy/AudioEncoder.java

@@ -37,6 +37,9 @@ public final class AudioEncoder implements AsyncProcessor {
     private static final int SAMPLE_RATE = AudioCapture.SAMPLE_RATE;
     private static final int CHANNELS = AudioCapture.CHANNELS;
 
+    private static final int READ_MS = 5; // milliseconds
+    private static final int READ_SIZE = AudioCapture.millisToBytes(READ_MS);
+
     private final AudioCapture capture;
     private final Streamer streamer;
     private final int bitRate;
@@ -90,7 +93,7 @@ public final class AudioEncoder implements AsyncProcessor {
         while (!Thread.currentThread().isInterrupted()) {
             InputTask task = inputTasks.take();
             ByteBuffer buffer = mediaCodec.getInputBuffer(task.index);
-            int r = capture.read(buffer, bufferInfo);
+            int r = capture.read(buffer, READ_SIZE, bufferInfo);
             if (r <= 0) {
                 throw new IOException("Could not read audio: " + r);
             }

server/src/main/java/com/genymobile/scrcpy/AudioRawRecorder.java

@@ -13,6 +13,9 @@ public final class AudioRawRecorder implements AsyncProcessor {
 
     private Thread thread;
 
+    private static final int READ_MS = 5; // milliseconds
+    private static final int READ_SIZE = AudioCapture.millisToBytes(READ_MS);
+
     public AudioRawRecorder(AudioCapture capture, Streamer streamer) {
         this.capture = capture;
         this.streamer = streamer;
@@ -25,7 +28,7 @@ public final class AudioRawRecorder implements AsyncProcessor {
             return;
         }
 
-        final ByteBuffer buffer = ByteBuffer.allocateDirect(AudioCapture.MAX_READ_SIZE);
+        final ByteBuffer buffer = ByteBuffer.allocateDirect(READ_SIZE);
         final MediaCodec.BufferInfo bufferInfo = new MediaCodec.BufferInfo();
 
         try {
@@ -40,7 +43,7 @@ public final class AudioRawRecorder implements AsyncProcessor {
             streamer.writeAudioHeader();
             while (!Thread.currentThread().isInterrupted()) {
                 buffer.position(0);
-                int r = capture.read(buffer, bufferInfo);
+                int r = capture.read(buffer, READ_SIZE, bufferInfo);
                 if (r < 0) {
                     throw new IOException("Could not read audio: " + r);
                 }