Encode Audio using Sink Writer - c++

I found this article that explaing how to encode a video using Media Foundation.
I am trying to encode an audio using the principle used in the above link.
I am stuck on setting a correct input media type for the sink writer.
Here is that part:
if (SUCCEEDED(hr))
hr = MFCreateMediaType(&pMediaTypeIn);
if (SUCCEEDED(hr))
hr = pMediaTypeIn->SetGUID(MF_MT_MAJOR_TYPE, MFMediaType_Audio);
if (SUCCEEDED(hr))
hr = pMediaTypeIn->SetGUID(MF_MT_SUBTYPE, MFAudioFormat_Float);
if (SUCCEEDED(hr))
hr = pMediaTypeIn->SetUINT32(MF_MT_AUDIO_NUM_CHANNELS, cChannels);
if (SUCCEEDED(hr))
hr = pMediaTypeIn->SetUINT32(MF_MT_AUDIO_SAMPLES_PER_SECOND, sampleRate);
if (SUCCEEDED(hr))
hr = pSinkWriter->SetInputMediaType(streamIndex, pMediaTypeIn, NULL);
My code fails on the last line while setting the input media type.
Any help, how to handle this?
Thanks.

You also need to supply MF_MT_AUDIO_BITS_PER_SAMPLE in the media type. It must be set to 16. Check the Input types requirements here: https://msdn.microsoft.com/en-us/library/windows/desktop/dd742785(v=vs.85).aspx. The sample rate must be either 44100 or 48000. The subtype must be MFAudioFormat_PCM.
The most important thing is that you need to call AddStream on the Sink Writer, prior to calling SetMediaType, with an output audio type enumerated by a call to MFTranscodeGetAudioOutputAvailableTypes function with MFAudioFormat_AAC subtype if you need to encode audio as AAC.
MFCreateSinkWriterFromURL may not be able to guess that you need to encode to an mp4 container from the m4a extension. You might need to supply MFTranscodeContainerType_MPEG4 containe type using the MF_TRANSCODE_CONTAINERTYPE attribute when calling MFCreateSinkWriterFromURL. Another option would be to use MFCreateMPEG4MediaSink and MFCreateSinkWriterFromMediaSink instead.

Related

UWP, Media Foundation, choosing specific encoder

I would like to choose a specific encoder in Media Foundation under UWP using c++/cx. Currently I use a SinkWriter and let the system choose a default encoder.
This code returns "class not registered" error under UWP, but it works in a win32 console app:
CoInitializeEx(NULL, COINIT_APARTMENTTHREADED | COINIT_DISABLE_OLE1DDE);
MFStartup(MF_VERSION);
IMFTransform* mtf;
CLSID id;
CLSIDFromString(L"{966F107C-8EA2-425D-B822-E4A71BEF01D7}", &id); // "NVIDIA HEVC Encoder MFT"
//CLSIDFromString(L"{F2F84074-8BCA-40BD-9159-E880F673DD3B}", &id); // "H265 Encoder MFT"
//CLSIDFromString(L"{BC10864D-2B34-408F-912A-102B1B867B6C}", &id); // "IntelĀ« Hardware H265 Encoder MFT"
//HRESULT hr = CoCreateInstance(id, nullptr, CLSCTX_INPROC_SERVER, IID_IMFTransform, (void **)&mtf);
HRESULT hr = CoCreateInstance(id, nullptr, CLSCTX_INPROC_SERVER, IID_PPV_ARGS(&mtf));
I also noticed that MFTEnumEx() is not definded in the header files under UWP, so I can't enumerate the encoders.
I noticed there is C# documentation allowing something like this:
auto codecQuery = ref new Windows::Media::Core::CodecQuery();
But it seems it not available when using c++/cx.
I would also like to ask the SinkWriter what encoder it actually chose, but this code does not work because ICodecAPI is undefined:
IMFTransform* pEncoder = NULL;
mWriter->GetServiceForStream(MF_SOURCE_READER_FIRST_VIDEO_STREAM, GUID_NULL, IID_IMFTransform, (void**)&pEncoder);
if (pEncoder)
{
ICodecAPI* pCodecApi = NULL;
hr = pEncoder->QueryInterface<ICodecAPI>(&pCodecApi);
}
Please help me choose encoder or find out which encoder was chosen?
Media Foundation does not offer flexibility to specify encoder using Sink Writer API. You can only instruct to use or not use hardware encoder, using MF_READWRITE_ENABLE_HARDWARE_TRANSFORMS attribute:
Enables the source reader or sink writer to use hardware-based Media Foundation transforms (MFTs).
Once Sink Writer is set up, you can use IMFSinkWriterEx::GetTransformForStream to enumerate the transforms the API prepared for the processing and pick the encoder from the enumeration. This is going to give you an idea what encoder is actually used.
Media Foundation Sink Writer API reserves the right to decode which encoder to use. Typically if would prefer certified compatible encoder, especially if you enable Direct3D scenario.
Finally, I am not sure which of these is available for C++/CX, but your code snippets suggest that the mentioned API is available.
To use encoder of your choice, you are supposed to use Media Foundation Media Session API, as opposed to Sink Writer.
Thank you Roman. I tried GetTranformForStream. With nvidia driver I get the attributes for the IMFTransform:
{206B4FC8-FCF9-4C51-AFE3-9764369E33A0}=1,
{2FB866AC-B078-4942-AB6C-003D05CDA674}=NVIDIA HEVC Encoder MFT,
FRIENDLY_NAME_Attribute=NVIDIA HEVC Encoder MFT,
{3AECB0CC-035B-4BCC-8185-2B8D551EF3AF}=VEN_10DE,
MAJOR_TYPE=Video,
{53476A11-3F13-49FB-AC42-EE2733C96741}=1,
{86A355AE-3A77-4EC4-9F31-01149A4E92DE}=1,
{88A7CB15-7B07-4A34-9128-E64C6703C4D3}=8,
{E3F2E203-D445-4B8C-9211-AE390D3BA017}=2303214,
{E5666D6B-3422-4EB6-A421-DA7DB1F8E207}=1,
{F34B9093-05E0-4B16-993D-3E2A2CDE6AD3}=860522,
SUBTYPE=Base,
{F81A699A-649A-497D-8C73-29F8FED6AD7A}=1,
When disabling nvidia driver I only get:
{86A355AE-3A77-4EC4-9F31-01149A4E92DE}=1
I wonder if the last transform is a list of several transforms? How to get them? Can I traverse the topology from sinkwriter?
My pc has the following codecs I could use:
{966F107C-8EA2-425D-B822-E4A71BEF01D7} // "NVIDIA HEVC Encoder MFT"
{F2F84074-8BCA-40BD-9159-E880F673DD3B} // "H265 Encoder MFT"
{BC10864D-2B34-408F-912A-102B1B867B6C} // "IntelĀ« Hardware H265 Encoder MFT"
In the nvidia case, I get a meaningful string, but not when it is not nvidia apparently (Intel or software).
Now I will also try look into Media Session API as you suggested.

Able to render video to MUX in GraphEdit but get VFW_E_CANNOT_CONNECT in code

I am trying to get the 3ivxfilters working in my C++ Directshow application and it continually fails to connect the 3IVX Video Encoder output pin to the 3IVX Media Muxer input pin. I always get the error VFW_E_CANNOT_CONNECT.
All filters have been added to the graph by enumerating the monikers so there shouldn't be any issue due to adding using a CLSID directly.
When I open the graph via graph edit and right click / choose render on the Video Encoder output pin it works fine.
Here is my code for connecting the filter:
HRESULT ConnectFilters(IGraphBuilder *pGraph, IBaseFilter *pSrc, IBaseFilter *pDest)
{
IPin *pOut = NULL;
// Find an output pin on the first filter.
HRESULT hr = FindUnconnectedPin(pSrc, PINDIR_OUTPUT, &pOut);
if (SUCCEEDED(hr))
{
hr = ConnectFilters(pGraph, pOut, pDest);
pOut->Release();
}
return hr;
}
Basically once it finds the appropriate pins it uses the Connect method.
hr = pGraph->Connect(pOut, pIn);

IMediaSample returned by Sample Grabber has unexpected buffer size

I was working on a audio/video capture library for Windows using Media Foundation. However, I faced the issue described in this post for some webcams on Windows 8.1. I have thus decided to have another implementation using Directshow to support in my app the webcams for which the drivers haven't been updated yet.
The library works quite well, but I have noticed a problem with some webcams for which the sample (IMediaSample) returned has not the expected size according to the format that was set before starting the camera.
For example, I have the case where the format set has a subtype of MEDIASUBTYPE_RGB24 (3 bytes per pixels) and the frame size is 640x480. The biSizeImage (from BITMAPINFOHEADER) is well 640*480*3 = 921600 when applying the format.
The IAMStreamConfig::SetFormat() method succeed to apply the format.
hr = pStreamConfig->SetFormat(pmt);
I also set the format to the Sample Grabber Interface as follow:
hr = pSampleGrabberInterface->SetMediaType(pmt);
I applied the format before starting the graph.
However, in the callback (ISampleGrabberCB::SampleCB), I'm receiving a sample of size 230400 (which might be a buffer for a frame of size 320x240 (320*240*3=230400)).
HRESULT MyClass::SampleCB(double sampleTime, IMediaSample *pSample)
{
unsigned char* pBuffer= 0;
HRESULT hr = pSample->GetPointer((unsigned char**) &pBuffer);
if(SUCCEEDED(hr) {
long bufSize = pSample->GetSize();
//bufSize = 230400
}
}
I tried to investigate the media type returned using the IMediaSample::GetMediaType() method, but the media type is NULL, which means, according to the documentation of the GetMediaType method that the media type has not changed (so I guess, it's still the media type I've applied successfully using the IAMStreamConfig::SetFormat() function).
HRESULT hr = pSample->GetMediaType(&pType);
if(SUCCEEDED(hr)) {
if(pType==NULL) {
//it enters here => the media type has not changed!
}
}
Why the sample buffer size returned is not the expected size in this case? How can I solve this issue?
Thanks in advance!
Sample Grabber callback will always return "correct" size in terms that it matches actual data size and formats used in streaming pipeline.
If you are seeing the mismatch, it means that your filter graph topology differs from what you expect it to be. You need to review the graph (esp. using remote connection by GraphEdit), inspect media types and check why it was built incorrectly. For example, you could be applying format of your interest after connecting pins, which is too late.
See also:
How can I reverse engineer a DirectShow graph?

Get encoder name from SinkWriter or ICodecAPI or IMFTransform

I'm using the SinkWriter in order to encode video using media foundation.
After I initialize the SinkWriter, I would like to get the underlying encoder it uses, and print out its name, so I can see what encoder it uses. (In my case, the encoder is most probably the H.264 Video Encoder included in MF).
I can get references to the encoder's ICodecAPI and IMFTransform interface (using pSinkWriter->GetServiceForStream), but I don't know how to get the encoder's friendly name using those interfaces.
Does anyone know how to get the encoder's friendly name from the sinkwriter? Or from its ICodecAPI or IMFTransform interface?
This is by far an effective solution and i am not 100% sure it works, but what could be done is:
1) At start-up enumerate all the codecs that could be used (as i understand in this case H264 encoders) and subscribe to setting change event
MFT_REGISTER_TYPE_INFO TransformationOutput = { MFMediaType_Video, MFVideoFormat_H264 };
DWORD nFlags = MFT_ENUM_FLAG_ALL;
UINT32 nCount = 0;
CLSID* pClsids;
MFTEnum( MFT_CATEGORY_VIDEO_ENCODER, nFlags, NULL, &TransformationOutput, NULL, &pClsids, &nCount);
// Ok here we assume nCount is 1 and we got the MS encoder
ICodecAPI *pMsEncoder;
hr = CoCreateInstance(pClsids[0], NULL, CLSCTX_INPROC_SERVER, __uuidof(ICodecAPI), (void**)&pMsEncoder);
// nCodecIds is supposed to be an array of identifiers to distinguish the sender
hr = pMsEncoder->RegisterForEvent(CODECAPI_AVEncVideoOutputFrameRate, (LONG_PTR)&nCodecIds[0]);
2) Not 100% sure if the frame rate setting is also set when the input media type for the stream is set, but anyhow you can try to set the same property on the ICodecAPI you retrieved from the SinkWriter. Then after getting the event you should be able to identify the codec by comparing lParam1 to the value passed. But still this is very poor since it relies on the fact that all the encoders support the event notification and requires unneeded parameter changing if my hypothesis about the event being generated on stream construction is wrong.
Having IMFTransform you don't have a friendly name of the encoder.
One of the options you have is to check transform output type and compare to well known GUIDs to identify the encoder, in particular you are going to have a subtype of MFVideoFormat_H264 with H264 Encoder MFT.
Another option is to reach CLSID of the encoder (IMFTransform does not get you it, but you might have it otherwise such as via IMFActivate or querying MFT_TRANSFORM_CLSID_Attribute attribute, or via IPersist* interfaces). Then you could look registry up for a friendly name or enumerate transforms and look your one in that list by comparing CLSID.

Sound from mic vs sound from speaker

I want to capture audio from both the mic and the speaker - separately. How can I distinguish between them? I can capture one or the other using the Wave API, e.g., WaveInOpen().
When I enumerate the devices using waveInGetNumDevs() and waveInGetDevCaps()/waveoutGetDevCaps(), there seems to be no information related to a particular end-point device (e.g., mic or speaker). I only see the following, which are adapter devices:
HD Read Audio Input
HD Read Audio Output
Webcam ...
I've actually no knowledge of the windows API so my answer isn't probably the best and there maybe even better ways.
HRESULT hr = CoInitialize(NULL);
IMMDeviceEnumerator *pEnum = NULL;
hr = CoCreateInstance(__uuidof(MMDeviceEnumerator), NULL, CLSCTX_ALL, __uuidof(IMMDeviceEnumerator), (void**)&pEnum);
if(SUCCEEDED(hr))
{
IMMDeviceCollection *pDevices;
// Enumerate the output devices.
hr = pEnum->EnumAudioEndpoints(eAll, DEVICE_STATE_ACTIVE, &pDevices);
// You can choose between eAll, eCapture or eRender
}
With that you'd be able to distinguish between input (capture) and output (render).
(That's what you wanted right?)
The code is taken from this article. You may look at it for the correct API calls and libraries, it even might give you some more information.
Hope that's helpfull.