Help needed for understanding of MPEG4-Part 17 (ISO/IEC 14496-17)

[Utkarsh]

I have gone through the MPEG4-Part 17 document and I have also gone through 3GPP TS 26.245 (Release 6) document.

Any help for the following queries would be very useful:

- Timed text track will have data format 'tx3g' in stsd. It will have TextConfig (defined in Part-17), but I was not able to find esds in any of the files. Is esds is mandatory?

- Part-17 tells about TTUs and Asset Units, but how the TTU are related to timed text track in .3gp or .mp4 file? Are TTUs in mdat or generated by some packatizer tool using text sample from mdat?

- What are in-band and out-band sample description indexes?

- What is content of sample description of TTU[5], it is contents of stsd?

- Any tool that can generate the .3gp or .mp4 file with timed text track.

- Any software or player that has implemented MPEG4 part-17

Thanks in Advance..

[Drachir]

...

- Any tool that can generate the .3gp or .mp4 file with timed text track.


You can use MP4Box from http://gpac.sf.net to create a 3gp file with 3GPP text stream. Take a look here: http://gpac.sourceforge.net/doc_ttxt.php

I am not sure if MP4 files (Usage of a text stream within an MPEG-4 system context) created by MP4Box are correct. As far as I remember is always "tx3g" used and not:
ISO/IEC FDIS 14496-17 2005
6.1 Signaling of a text stream
When used in an MPEG-4 system context, a text stream shall be signaled by a streamType value 0x0D and by an objectTypeIndication value of 0x08.

But I am not up to date about 3GPP text stream in MPEG-4 system context.

You might find some help here http://sourceforge.net/forum/forum.php?forum_id=327349 or here http://www.m4if.org/public/publiclistreg.php .

[Utkarsh]

Thanks for the reply...

I have used MP4Box to generate .3gp and .mp4 file with timed text. I have following observation and understading. Please correct me If I am wrong.

I seen the content of the file and parsed stsd box, it seems that, it follows 3GPP TS 26.245 standard. It does not contain esds. So, text stream is not signaled by streamType and objectTypeIndication, but MPEG-4 Part-17 tells it should be.

I have also parsed mdat. The sample starts with 2 byte length of text string followed by text string. The text sample may contain text modifiers. The format of text sample are not as per MPEG-4 Part 17. Text samples are not made up of TTUs. Text samples are as per 3GPP TS 26.245 standard.

I think Part-17 tells that how to stream timed text which is signaled as 'tx3g' (Timed Text), it doesn't tell how the text sample is organized in the 3GP or MP4 file. It doesn't tell how to interpret data in mdat and text sample. But it tells how to stream text sample. So, I think TTUs are not present in 3GP or MP4 file, but generated by some packetizer tool. The packetizer tool gets the offset and size of text sample from moov and read the text sample. The text sample is made up of text string and text modifiers. The packetizer tool makes TTUs out of these text samples and streams it over the network. In brief, 3GPP TS 26.245 tells how to write/read timed text track in 3GP or MP4 file, how the sample should be organized. MPEG-4 Part-17 tells how to stream the packets. MPEG-4 Part-17 tells how the packetizer tool should generate Access Units which are made up of TTUs using the text sample from the 3GP or MP4 file. It does not give any information about file format except TextConfig. This is my understanding, but I don't know whether it is correct or not.

Waiting for the reply.. Thanks in Advance..