Imported Debian version 2.5.0~trusty1.1
[deb_ffmpeg.git] / ffmpeg / libavdevice / avfoundation.m
index 4ac50a0ef2f9e8a17661358b305c55a6c16739df..1a7eb5b27e431ac306a9f52c8983e26568f6f8e4 100644 (file)
 
 #include "libavutil/pixdesc.h"
 #include "libavutil/opt.h"
+#include "libavutil/avstring.h"
 #include "libavformat/internal.h"
 #include "libavutil/internal.h"
 #include "libavutil/time.h"
 #include "avdevice.h"
 
-static const int avf_time_base = 100;
+static const int avf_time_base = 1000000;
 
 static const AVRational avf_time_base_q = {
     .num = 1,
@@ -80,20 +81,44 @@ typedef struct
 {
     AVClass*        class;
 
-    float           frame_rate;
     int             frames_captured;
+    int             audio_frames_captured;
     int64_t         first_pts;
+    int64_t         first_audio_pts;
     pthread_mutex_t frame_lock;
     pthread_cond_t  frame_wait_cond;
     id              avf_delegate;
+    id              avf_audio_delegate;
 
     int             list_devices;
     int             video_device_index;
+    int             video_stream_index;
+    int             audio_device_index;
+    int             audio_stream_index;
+
+    char            *video_filename;
+    char            *audio_filename;
+
+    int             num_video_devices;
+
+    int             audio_channels;
+    int             audio_bits_per_sample;
+    int             audio_float;
+    int             audio_be;
+    int             audio_signed_integer;
+    int             audio_packed;
+    int             audio_non_interleaved;
+
+    int32_t         *audio_buffer;
+    int             audio_buffer_size;
+
     enum AVPixelFormat pixel_format;
 
     AVCaptureSession         *capture_session;
     AVCaptureVideoDataOutput *video_output;
+    AVCaptureAudioDataOutput *audio_output;
     CMSampleBufferRef         current_frame;
+    CMSampleBufferRef         current_audio_frame;
 } AVFContext;
 
 static void lock_frames(AVFContext* ctx)
@@ -152,17 +177,69 @@ static void unlock_frames(AVFContext* ctx)
 
 @end
 
+/** AudioReciever class - delegate for AVCaptureSession
+ */
+@interface AVFAudioReceiver : NSObject
+{
+    AVFContext* _context;
+}
+
+- (id)initWithContext:(AVFContext*)context;
+
+- (void)  captureOutput:(AVCaptureOutput *)captureOutput
+  didOutputSampleBuffer:(CMSampleBufferRef)audioFrame
+         fromConnection:(AVCaptureConnection *)connection;
+
+@end
+
+@implementation AVFAudioReceiver
+
+- (id)initWithContext:(AVFContext*)context
+{
+    if (self = [super init]) {
+        _context = context;
+    }
+    return self;
+}
+
+- (void)  captureOutput:(AVCaptureOutput *)captureOutput
+  didOutputSampleBuffer:(CMSampleBufferRef)audioFrame
+         fromConnection:(AVCaptureConnection *)connection
+{
+    lock_frames(_context);
+
+    if (_context->current_audio_frame != nil) {
+        CFRelease(_context->current_audio_frame);
+    }
+
+    _context->current_audio_frame = (CMSampleBufferRef)CFRetain(audioFrame);
+
+    pthread_cond_signal(&_context->frame_wait_cond);
+
+    unlock_frames(_context);
+
+    ++_context->audio_frames_captured;
+}
+
+@end
+
 static void destroy_context(AVFContext* ctx)
 {
     [ctx->capture_session stopRunning];
 
     [ctx->capture_session release];
     [ctx->video_output    release];
+    [ctx->audio_output    release];
     [ctx->avf_delegate    release];
+    [ctx->avf_audio_delegate release];
 
     ctx->capture_session = NULL;
     ctx->video_output    = NULL;
+    ctx->audio_output    = NULL;
     ctx->avf_delegate    = NULL;
+    ctx->avf_audio_delegate = NULL;
+
+    av_freep(&ctx->audio_buffer);
 
     pthread_mutex_destroy(&ctx->frame_lock);
     pthread_cond_destroy(&ctx->frame_wait_cond);
@@ -172,99 +249,43 @@ static void destroy_context(AVFContext* ctx)
     }
 }
 
-static int avf_read_header(AVFormatContext *s)
+static void parse_device_name(AVFormatContext *s)
 {
-    NSAutoreleasePool *pool = [[NSAutoreleasePool alloc] init];
-    AVFContext *ctx         = (AVFContext*)s->priv_data;
-    ctx->first_pts          = av_gettime();
-
-    pthread_mutex_init(&ctx->frame_lock, NULL);
-    pthread_cond_init(&ctx->frame_wait_cond, NULL);
-
-    // List devices if requested
-    if (ctx->list_devices) {
-        av_log(ctx, AV_LOG_INFO, "AVFoundation video devices:\n");
-        NSArray *devices = [AVCaptureDevice devicesWithMediaType:AVMediaTypeVideo];
-        for (AVCaptureDevice *device in devices) {
-            const char *name = [[device localizedName] UTF8String];
-            int index  = [devices indexOfObject:device];
-            av_log(ctx, AV_LOG_INFO, "[%d] %s\n", index, name);
-        }
-        goto fail;
-    }
-
-    // Find capture device
-    AVCaptureDevice *video_device = nil;
-
-    // check for device index given in filename
-    if (ctx->video_device_index == -1) {
-        sscanf(s->filename, "%d", &ctx->video_device_index);
-    }
-
-    if (ctx->video_device_index >= 0) {
-        NSArray *devices = [AVCaptureDevice devicesWithMediaType:AVMediaTypeVideo];
-
-        if (ctx->video_device_index >= [devices count]) {
-            av_log(ctx, AV_LOG_ERROR, "Invalid device index\n");
-            goto fail;
-        }
-
-        video_device = [devices objectAtIndex:ctx->video_device_index];
-    } else if (strncmp(s->filename, "",        1) &&
-               strncmp(s->filename, "default", 7)) {
-        NSArray *devices = [AVCaptureDevice devicesWithMediaType:AVMediaTypeVideo];
-
-        for (AVCaptureDevice *device in devices) {
-            if (!strncmp(s->filename, [[device localizedName] UTF8String], strlen(s->filename))) {
-                video_device = device;
-                break;
-            }
-        }
+    AVFContext *ctx = (AVFContext*)s->priv_data;
+    char *tmp = av_strdup(s->filename);
+    char *save;
 
-        if (!video_device) {
-            av_log(ctx, AV_LOG_ERROR, "Video device not found\n");
-            goto fail;
-        }
+    if (tmp[0] != ':') {
+        ctx->video_filename = av_strtok(tmp,  ":", &save);
+        ctx->audio_filename = av_strtok(NULL, ":", &save);
     } else {
-        video_device = [AVCaptureDevice defaultDeviceWithMediaType:AVMediaTypeMuxed];
+        ctx->audio_filename = av_strtok(tmp,  ":", &save);
     }
+}
 
-    // Video capture device not found, looking for AVMediaTypeVideo
-    if (!video_device) {
-        video_device = [AVCaptureDevice defaultDeviceWithMediaType:AVMediaTypeVideo];
+static int add_video_device(AVFormatContext *s, AVCaptureDevice *video_device)
+{
+    AVFContext *ctx = (AVFContext*)s->priv_data;
+    NSError *error  = nil;
+    AVCaptureInput* capture_input = nil;
 
-        if (!video_device) {
-            av_log(s, AV_LOG_ERROR, "No AV capture device found\n");
-            goto fail;
-        }
+    if (ctx->video_device_index < ctx->num_video_devices) {
+        capture_input = (AVCaptureInput*) [[[AVCaptureDeviceInput alloc] initWithDevice:video_device error:&error] autorelease];
+    } else {
+        capture_input = (AVCaptureInput*) video_device;
     }
 
-    NSString* dev_display_name = [video_device localizedName];
-    av_log(s, AV_LOG_DEBUG, "'%s' opened\n", [dev_display_name UTF8String]);
-
-    // Initialize capture session
-    ctx->capture_session = [[AVCaptureSession alloc] init];
-
-    NSError *error = nil;
-    AVCaptureDeviceInput* capture_dev_input = [[[AVCaptureDeviceInput alloc] initWithDevice:video_device error:&error] autorelease];
-
-    if (!capture_dev_input) {
+    if (!capture_input) {
         av_log(s, AV_LOG_ERROR, "Failed to create AV capture input device: %s\n",
                [[error localizedDescription] UTF8String]);
-        goto fail;
-    }
-
-    if (!capture_dev_input) {
-        av_log(s, AV_LOG_ERROR, "Failed to add AV capture input device to session: %s\n",
-               [[error localizedDescription] UTF8String]);
-        goto fail;
+        return 1;
     }
 
-    if ([ctx->capture_session canAddInput:capture_dev_input]) {
-        [ctx->capture_session addInput:capture_dev_input];
+    if ([ctx->capture_session canAddInput:capture_input]) {
+        [ctx->capture_session addInput:capture_input];
     } else {
         av_log(s, AV_LOG_ERROR, "can't add video input to capture session\n");
-        goto fail;
+        return 1;
     }
 
     // Attaching output
@@ -272,7 +293,7 @@ static int avf_read_header(AVFormatContext *s)
 
     if (!ctx->video_output) {
         av_log(s, AV_LOG_ERROR, "Failed to init AV video output\n");
-        goto fail;
+        return 1;
     }
 
     // select pixel format
@@ -290,7 +311,7 @@ static int avf_read_header(AVFormatContext *s)
     if (pxl_fmt_spec.ff_id == AV_PIX_FMT_NONE) {
         av_log(s, AV_LOG_ERROR, "Selected pixel format (%s) is not supported by AVFoundation.\n",
                av_get_pix_fmt_name(pxl_fmt_spec.ff_id));
-        goto fail;
+        return 1;
     }
 
     // check if the pixel format is available for this device
@@ -323,14 +344,14 @@ static int avf_read_header(AVFormatContext *s)
 
         // fail if there is no appropriate pixel format or print a warning about overriding the pixel format
         if (pxl_fmt_spec.ff_id == AV_PIX_FMT_NONE) {
-            goto fail;
+            return 1;
         } else {
             av_log(s, AV_LOG_WARNING, "Overriding selected pixel format to use %s instead.\n",
                    av_get_pix_fmt_name(pxl_fmt_spec.ff_id));
         }
     }
 
-
+    ctx->pixel_format          = pxl_fmt_spec.ff_id;
     NSNumber     *pixel_format = [NSNumber numberWithUnsignedInt:pxl_fmt_spec.avf_id];
     NSDictionary *capture_dict = [NSDictionary dictionaryWithObject:pixel_format
                                                forKey:(id)kCVPixelBufferPixelFormatTypeKey];
@@ -348,10 +369,58 @@ static int avf_read_header(AVFormatContext *s)
         [ctx->capture_session addOutput:ctx->video_output];
     } else {
         av_log(s, AV_LOG_ERROR, "can't add video output to capture session\n");
-        goto fail;
+        return 1;
     }
 
-    [ctx->capture_session startRunning];
+    return 0;
+}
+
+static int add_audio_device(AVFormatContext *s, AVCaptureDevice *audio_device)
+{
+    AVFContext *ctx = (AVFContext*)s->priv_data;
+    NSError *error  = nil;
+    AVCaptureDeviceInput* audio_dev_input = [[[AVCaptureDeviceInput alloc] initWithDevice:audio_device error:&error] autorelease];
+
+    if (!audio_dev_input) {
+        av_log(s, AV_LOG_ERROR, "Failed to create AV capture input device: %s\n",
+               [[error localizedDescription] UTF8String]);
+        return 1;
+    }
+
+    if ([ctx->capture_session canAddInput:audio_dev_input]) {
+        [ctx->capture_session addInput:audio_dev_input];
+    } else {
+        av_log(s, AV_LOG_ERROR, "can't add audio input to capture session\n");
+        return 1;
+    }
+
+    // Attaching output
+    ctx->audio_output = [[AVCaptureAudioDataOutput alloc] init];
+
+    if (!ctx->audio_output) {
+        av_log(s, AV_LOG_ERROR, "Failed to init AV audio output\n");
+        return 1;
+    }
+
+    ctx->avf_audio_delegate = [[AVFAudioReceiver alloc] initWithContext:ctx];
+
+    dispatch_queue_t queue = dispatch_queue_create("avf_audio_queue", NULL);
+    [ctx->audio_output setSampleBufferDelegate:ctx->avf_audio_delegate queue:queue];
+    dispatch_release(queue);
+
+    if ([ctx->capture_session canAddOutput:ctx->audio_output]) {
+        [ctx->capture_session addOutput:ctx->audio_output];
+    } else {
+        av_log(s, AV_LOG_ERROR, "adding audio output to capture session failed\n");
+        return 1;
+    }
+
+    return 0;
+}
+
+static int get_video_config(AVFormatContext *s)
+{
+    AVFContext *ctx = (AVFContext*)s->priv_data;
 
     // Take stream info from the first frame.
     while (ctx->frames_captured < 1) {
@@ -363,9 +432,11 @@ static int avf_read_header(AVFormatContext *s)
     AVStream* stream = avformat_new_stream(s, NULL);
 
     if (!stream) {
-        goto fail;
+        return 1;
     }
 
+    ctx->video_stream_index = stream->index;
+
     avpriv_set_pts_info(stream, 64, 1, avf_time_base);
 
     CVImageBufferRef image_buffer = CMSampleBufferGetImageBuffer(ctx->current_frame);
@@ -375,12 +446,265 @@ static int avf_read_header(AVFormatContext *s)
     stream->codec->codec_type = AVMEDIA_TYPE_VIDEO;
     stream->codec->width      = (int)image_buffer_size.width;
     stream->codec->height     = (int)image_buffer_size.height;
-    stream->codec->pix_fmt    = pxl_fmt_spec.ff_id;
+    stream->codec->pix_fmt    = ctx->pixel_format;
 
     CFRelease(ctx->current_frame);
     ctx->current_frame = nil;
 
     unlock_frames(ctx);
+
+    return 0;
+}
+
+static int get_audio_config(AVFormatContext *s)
+{
+    AVFContext *ctx = (AVFContext*)s->priv_data;
+
+    // Take stream info from the first frame.
+    while (ctx->audio_frames_captured < 1) {
+        CFRunLoopRunInMode(kCFRunLoopDefaultMode, 0.1, YES);
+    }
+
+    lock_frames(ctx);
+
+    AVStream* stream = avformat_new_stream(s, NULL);
+
+    if (!stream) {
+        return 1;
+    }
+
+    ctx->audio_stream_index = stream->index;
+
+    avpriv_set_pts_info(stream, 64, 1, avf_time_base);
+
+    CMFormatDescriptionRef format_desc = CMSampleBufferGetFormatDescription(ctx->current_audio_frame);
+    const AudioStreamBasicDescription *basic_desc = CMAudioFormatDescriptionGetStreamBasicDescription(format_desc);
+
+    if (!basic_desc) {
+        av_log(s, AV_LOG_ERROR, "audio format not available\n");
+        return 1;
+    }
+
+    stream->codec->codec_type     = AVMEDIA_TYPE_AUDIO;
+    stream->codec->sample_rate    = basic_desc->mSampleRate;
+    stream->codec->channels       = basic_desc->mChannelsPerFrame;
+    stream->codec->channel_layout = av_get_default_channel_layout(stream->codec->channels);
+
+    ctx->audio_channels        = basic_desc->mChannelsPerFrame;
+    ctx->audio_bits_per_sample = basic_desc->mBitsPerChannel;
+    ctx->audio_float           = basic_desc->mFormatFlags & kAudioFormatFlagIsFloat;
+    ctx->audio_be              = basic_desc->mFormatFlags & kAudioFormatFlagIsBigEndian;
+    ctx->audio_signed_integer  = basic_desc->mFormatFlags & kAudioFormatFlagIsSignedInteger;
+    ctx->audio_packed          = basic_desc->mFormatFlags & kAudioFormatFlagIsPacked;
+    ctx->audio_non_interleaved = basic_desc->mFormatFlags & kAudioFormatFlagIsNonInterleaved;
+
+    if (basic_desc->mFormatID == kAudioFormatLinearPCM &&
+        ctx->audio_float &&
+        ctx->audio_packed) {
+        stream->codec->codec_id = ctx->audio_be ? AV_CODEC_ID_PCM_F32BE : AV_CODEC_ID_PCM_F32LE;
+    } else {
+        av_log(s, AV_LOG_ERROR, "audio format is not supported\n");
+        return 1;
+    }
+
+    if (ctx->audio_non_interleaved) {
+        CMBlockBufferRef block_buffer = CMSampleBufferGetDataBuffer(ctx->current_audio_frame);
+        ctx->audio_buffer_size        = CMBlockBufferGetDataLength(block_buffer);
+        ctx->audio_buffer             = av_malloc(ctx->audio_buffer_size);
+        if (!ctx->audio_buffer) {
+            av_log(s, AV_LOG_ERROR, "error allocating audio buffer\n");
+            return 1;
+        }
+    }
+
+    CFRelease(ctx->current_audio_frame);
+    ctx->current_audio_frame = nil;
+
+    unlock_frames(ctx);
+
+    return 0;
+}
+
+static int avf_read_header(AVFormatContext *s)
+{
+    NSAutoreleasePool *pool = [[NSAutoreleasePool alloc] init];
+    AVFContext *ctx         = (AVFContext*)s->priv_data;
+    ctx->first_pts          = av_gettime();
+    ctx->first_audio_pts    = av_gettime();
+    uint32_t num_screens    = 0;
+
+    pthread_mutex_init(&ctx->frame_lock, NULL);
+    pthread_cond_init(&ctx->frame_wait_cond, NULL);
+
+#if __MAC_OS_X_VERSION_MIN_REQUIRED >= 1070
+    CGGetActiveDisplayList(0, NULL, &num_screens);
+#endif
+
+    // List devices if requested
+    if (ctx->list_devices) {
+        av_log(ctx, AV_LOG_INFO, "AVFoundation video devices:\n");
+        NSArray *devices = [AVCaptureDevice devicesWithMediaType:AVMediaTypeVideo];
+        int index = 0;
+        for (AVCaptureDevice *device in devices) {
+            const char *name = [[device localizedName] UTF8String];
+            index            = [devices indexOfObject:device];
+            av_log(ctx, AV_LOG_INFO, "[%d] %s\n", index, name);
+            index++;
+        }
+#if __MAC_OS_X_VERSION_MIN_REQUIRED >= 1070
+        if (num_screens > 0) {
+            CGDirectDisplayID screens[num_screens];
+            CGGetActiveDisplayList(num_screens, screens, &num_screens);
+            for (int i = 0; i < num_screens; i++) {
+                av_log(ctx, AV_LOG_INFO, "[%d] Capture screen %d\n", index + i, i);
+            }
+        }
+#endif
+
+        av_log(ctx, AV_LOG_INFO, "AVFoundation audio devices:\n");
+        devices = [AVCaptureDevice devicesWithMediaType:AVMediaTypeAudio];
+        for (AVCaptureDevice *device in devices) {
+            const char *name = [[device localizedName] UTF8String];
+            int index  = [devices indexOfObject:device];
+            av_log(ctx, AV_LOG_INFO, "[%d] %s\n", index, name);
+        }
+         goto fail;
+    }
+
+    // Find capture device
+    AVCaptureDevice *video_device = nil;
+    AVCaptureDevice *audio_device = nil;
+
+    NSArray *video_devices = [AVCaptureDevice devicesWithMediaType:AVMediaTypeVideo];
+    ctx->num_video_devices = [video_devices count];
+
+    // parse input filename for video and audio device
+    parse_device_name(s);
+
+    // check for device index given in filename
+    if (ctx->video_device_index == -1 && ctx->video_filename) {
+        sscanf(ctx->video_filename, "%d", &ctx->video_device_index);
+    }
+    if (ctx->audio_device_index == -1 && ctx->audio_filename) {
+        sscanf(ctx->audio_filename, "%d", &ctx->audio_device_index);
+    }
+
+    if (ctx->video_device_index >= 0) {
+        if (ctx->video_device_index < ctx->num_video_devices) {
+            video_device = [video_devices objectAtIndex:ctx->video_device_index];
+        } else if (ctx->video_device_index < ctx->num_video_devices + num_screens) {
+#if __MAC_OS_X_VERSION_MIN_REQUIRED >= 1070
+            CGDirectDisplayID screens[num_screens];
+            CGGetActiveDisplayList(num_screens, screens, &num_screens);
+            AVCaptureScreenInput* capture_screen_input = [[[AVCaptureScreenInput alloc] initWithDisplayID:screens[ctx->video_device_index - ctx->num_video_devices]] autorelease];
+            video_device = (AVCaptureDevice*) capture_screen_input;
+#endif
+         } else {
+            av_log(ctx, AV_LOG_ERROR, "Invalid device index\n");
+            goto fail;
+        }
+    } else if (ctx->video_filename &&
+               strncmp(ctx->video_filename, "none", 4)) {
+        if (!strncmp(ctx->video_filename, "default", 7)) {
+            video_device = [AVCaptureDevice defaultDeviceWithMediaType:AVMediaTypeVideo];
+        } else {
+        // looking for video inputs
+        for (AVCaptureDevice *device in video_devices) {
+            if (!strncmp(ctx->video_filename, [[device localizedName] UTF8String], strlen(ctx->video_filename))) {
+                video_device = device;
+                break;
+            }
+        }
+
+#if __MAC_OS_X_VERSION_MIN_REQUIRED >= 1070
+        // looking for screen inputs
+        if (!video_device) {
+            int idx;
+            if(sscanf(ctx->video_filename, "Capture screen %d", &idx) && idx < num_screens) {
+                CGDirectDisplayID screens[num_screens];
+                CGGetActiveDisplayList(num_screens, screens, &num_screens);
+                AVCaptureScreenInput* capture_screen_input = [[[AVCaptureScreenInput alloc] initWithDisplayID:screens[idx]] autorelease];
+                video_device = (AVCaptureDevice*) capture_screen_input;
+                ctx->video_device_index = ctx->num_video_devices + idx;
+            }
+        }
+#endif
+        }
+
+        if (!video_device) {
+            av_log(ctx, AV_LOG_ERROR, "Video device not found\n");
+            goto fail;
+        }
+    }
+
+    // get audio device
+    if (ctx->audio_device_index >= 0) {
+        NSArray *devices = [AVCaptureDevice devicesWithMediaType:AVMediaTypeAudio];
+
+        if (ctx->audio_device_index >= [devices count]) {
+            av_log(ctx, AV_LOG_ERROR, "Invalid audio device index\n");
+            goto fail;
+        }
+
+        audio_device = [devices objectAtIndex:ctx->audio_device_index];
+    } else if (ctx->audio_filename &&
+               strncmp(ctx->audio_filename, "none", 4)) {
+        if (!strncmp(ctx->audio_filename, "default", 7)) {
+            audio_device = [AVCaptureDevice defaultDeviceWithMediaType:AVMediaTypeAudio];
+        } else {
+        NSArray *devices = [AVCaptureDevice devicesWithMediaType:AVMediaTypeAudio];
+
+        for (AVCaptureDevice *device in devices) {
+            if (!strncmp(ctx->audio_filename, [[device localizedName] UTF8String], strlen(ctx->audio_filename))) {
+                audio_device = device;
+                break;
+            }
+        }
+        }
+
+        if (!audio_device) {
+            av_log(ctx, AV_LOG_ERROR, "Audio device not found\n");
+             goto fail;
+        }
+    }
+
+    // Video nor Audio capture device not found, looking for AVMediaTypeVideo/Audio
+    if (!video_device && !audio_device) {
+        av_log(s, AV_LOG_ERROR, "No AV capture device found\n");
+        goto fail;
+    }
+
+    if (video_device) {
+        if (ctx->video_device_index < ctx->num_video_devices) {
+            av_log(s, AV_LOG_DEBUG, "'%s' opened\n", [[video_device localizedName] UTF8String]);
+        } else {
+            av_log(s, AV_LOG_DEBUG, "'%s' opened\n", [[video_device description] UTF8String]);
+        }
+    }
+    if (audio_device) {
+        av_log(s, AV_LOG_DEBUG, "audio device '%s' opened\n", [[audio_device localizedName] UTF8String]);
+    }
+
+    // Initialize capture session
+    ctx->capture_session = [[AVCaptureSession alloc] init];
+
+    if (video_device && add_video_device(s, video_device)) {
+        goto fail;
+    }
+    if (audio_device && add_audio_device(s, audio_device)) {
+    }
+
+    [ctx->capture_session startRunning];
+
+    if (video_device && get_video_config(s)) {
+        goto fail;
+    }
+
+    // set audio stream
+    if (audio_device && get_audio_config(s)) {
+        goto fail;
+    }
+
     [pool release];
     return 0;
 
@@ -407,7 +731,7 @@ static int avf_read_packet(AVFormatContext *s, AVPacket *pkt)
             pkt->pts = pkt->dts = av_rescale_q(av_gettime() - ctx->first_pts,
                                                AV_TIME_BASE_Q,
                                                avf_time_base_q);
-            pkt->stream_index  = 0;
+            pkt->stream_index  = ctx->video_stream_index;
             pkt->flags        |= AV_PKT_FLAG_KEY;
 
             CVPixelBufferLockBaseAddress(image_buffer, 0);
@@ -418,6 +742,71 @@ static int avf_read_packet(AVFormatContext *s, AVPacket *pkt)
             CVPixelBufferUnlockBaseAddress(image_buffer, 0);
             CFRelease(ctx->current_frame);
             ctx->current_frame = nil;
+        } else if (ctx->current_audio_frame != nil) {
+            CMBlockBufferRef block_buffer = CMSampleBufferGetDataBuffer(ctx->current_audio_frame);
+            int block_buffer_size         = CMBlockBufferGetDataLength(block_buffer);
+
+            if (!block_buffer || !block_buffer_size) {
+                return AVERROR(EIO);
+            }
+
+            if (ctx->audio_non_interleaved && block_buffer_size > ctx->audio_buffer_size) {
+                return AVERROR_BUFFER_TOO_SMALL;
+            }
+
+            if (av_new_packet(pkt, block_buffer_size) < 0) {
+                return AVERROR(EIO);
+            }
+
+            pkt->pts = pkt->dts = av_rescale_q(av_gettime() - ctx->first_audio_pts,
+                                               AV_TIME_BASE_Q,
+                                               avf_time_base_q);
+
+            pkt->stream_index  = ctx->audio_stream_index;
+            pkt->flags        |= AV_PKT_FLAG_KEY;
+
+            if (ctx->audio_non_interleaved) {
+                int sample, c, shift;
+
+                OSStatus ret = CMBlockBufferCopyDataBytes(block_buffer, 0, pkt->size, ctx->audio_buffer);
+                if (ret != kCMBlockBufferNoErr) {
+                    return AVERROR(EIO);
+                }
+
+                int num_samples = pkt->size / (ctx->audio_channels * (ctx->audio_bits_per_sample >> 3));
+
+                // transform decoded frame into output format
+                #define INTERLEAVE_OUTPUT(bps)                                         \
+                {                                                                      \
+                    int##bps##_t **src;                                                \
+                    int##bps##_t *dest;                                                \
+                    src = av_malloc(ctx->audio_channels * sizeof(int##bps##_t*));      \
+                    if (!src) return AVERROR(EIO);                                     \
+                    for (c = 0; c < ctx->audio_channels; c++) {                        \
+                        src[c] = ((int##bps##_t*)ctx->audio_buffer) + c * num_samples; \
+                    }                                                                  \
+                    dest  = (int##bps##_t*)pkt->data;                                  \
+                    shift = bps - ctx->audio_bits_per_sample;                          \
+                    for (sample = 0; sample < num_samples; sample++)                   \
+                        for (c = 0; c < ctx->audio_channels; c++)                      \
+                            *dest++ = src[c][sample] << shift;                         \
+                    av_freep(&src);                                                    \
+                }
+
+                if (ctx->audio_bits_per_sample <= 16) {
+                    INTERLEAVE_OUTPUT(16)
+                } else {
+                    INTERLEAVE_OUTPUT(32)
+                }
+            } else {
+                OSStatus ret = CMBlockBufferCopyDataBytes(block_buffer, 0, pkt->size, pkt->data);
+                if (ret != kCMBlockBufferNoErr) {
+                    return AVERROR(EIO);
+                }
+            }
+
+            CFRelease(ctx->current_audio_frame);
+            ctx->current_audio_frame = nil;
         } else {
             pkt->data = NULL;
             pthread_cond_wait(&ctx->frame_wait_cond, &ctx->frame_lock);
@@ -437,11 +826,11 @@ static int avf_close(AVFormatContext *s)
 }
 
 static const AVOption options[] = {
-    { "frame_rate", "set frame rate", offsetof(AVFContext, frame_rate), AV_OPT_TYPE_FLOAT, { .dbl = 30.0 }, 0.1, 30.0, AV_OPT_TYPE_VIDEO_RATE, NULL },
     { "list_devices", "list available devices", offsetof(AVFContext, list_devices), AV_OPT_TYPE_INT, {.i64=0}, 0, 1, AV_OPT_FLAG_DECODING_PARAM, "list_devices" },
     { "true", "", 0, AV_OPT_TYPE_CONST, {.i64=1}, 0, 0, AV_OPT_FLAG_DECODING_PARAM, "list_devices" },
     { "false", "", 0, AV_OPT_TYPE_CONST, {.i64=0}, 0, 0, AV_OPT_FLAG_DECODING_PARAM, "list_devices" },
     { "video_device_index", "select video device by index for devices with same name (starts at 0)", offsetof(AVFContext, video_device_index), AV_OPT_TYPE_INT, {.i64 = -1}, -1, INT_MAX, AV_OPT_FLAG_DECODING_PARAM },
+    { "audio_device_index", "select audio device by index for devices with same name (starts at 0)", offsetof(AVFContext, audio_device_index), AV_OPT_TYPE_INT, {.i64 = -1}, -1, INT_MAX, AV_OPT_FLAG_DECODING_PARAM },
     { "pixel_format", "set pixel format", offsetof(AVFContext, pixel_format), AV_OPT_TYPE_PIXEL_FMT, {.i64 = AV_PIX_FMT_YUV420P}, 0, INT_MAX, AV_OPT_FLAG_DECODING_PARAM},
     { NULL },
 };