I have to transcode with FFMpeg some video with multiple audio where various audio track are not language related, just music and so on.
I would like to assing as language something like Unspecified, Unknown, Not Applicable, Multilanguage, or so on.
Everything that is not a specific named language (English, French, Spanish, ...) works fine for me .
ffmpeg -i in.mov -map 0:a -metadata:s:a:0 language="und" -metadata:s:a:0 title="Just Music" out.mov
For what I researched FFMPeg uses the ISO 639-2 codes:
https://en.wikipedia.org/wiki/List_of_ISO_639-2_codes
In my example I tried to use the und that stands for Undetermined..
In the wikipedia list there are also other possible interesting options:
The standard includes some codes for special situations:
mis, for "uncoded languages";
mul, for "multiple languages";
qaa-qtz, a range reserved for local use.
und, for "undetermined";
zxx, for "no linguistic content, not applicable";
Non of them seems to work, everytime I try to set that values with ffmpeg it just assume "English".
Following the Apple Documentation for Moov Format we have:
A very small list compared to the Wikipedia one but there's at least a Unspecified 32767
I tried with -metadata:s:a:0 language="und"
or -metadata:s:a:0 language="32767"
but is not working the same :(
So I tried to read the source code of ffmpeg, at ffmpeg/libavformat/isom.c line 382:
/* map numeric codes from mdhd atom to ISO 639 */
/* cf. QTFileFormat.pdf p253, qtff.pdf p205 */
/* http://developer.apple.com/documentation/mac/Text/Text-368.html */
/* deprecated by putting the code as 3*5 bits ASCII */
static const char mov_mdhd_language_map[][4] = {
/* 0-9 */
"eng", "fra", "ger", "ita", "dut", "sve", "spa", "dan", "por", "nor",
"heb", "jpn", "ara", "fin", "gre", "ice", "mlt", "tur", "hr "/*scr*/, "chi"/*ace?*/,
"urd", "hin", "tha", "kor", "lit", "pol", "hun", "est", "lav", "",
"fo ", "", "rus", "chi", "", "iri", "alb", "ron", "ces", "slk",
"slv", "yid", "sr ", "mac", "bul", "ukr", "bel", "uzb", "kaz", "aze",
/*?*/
"aze", "arm", "geo", "mol", "kir", "tgk", "tuk", "mon", "", "pus",
"kur", "kas", "snd", "tib", "nep", "san", "mar", "ben", "asm", "guj",
"pa ", "ori", "mal", "kan", "tam", "tel", "", "bur", "khm", "lao",
/* roman? arabic? */
"vie", "ind", "tgl", "may", "may", "amh", "tir", "orm", "som", "swa",
/*==rundi?*/
"", "run", "", "mlg", "epo", "", "", "", "", "",
/* 100 */
"", "", "", "", "", "", "", "", "", "",
"", "", "", "", "", "", "", "", "", "",
"", "", "", "", "", "", "", "", "wel", "baq",
"cat", "lat", "que", "grn", "aym", "tat", "uig", "dzo", "jav"
};
int ff_mov_iso639_to_lang(const char lang[4], int mp4)
{
int i, code = 0;
/* old way, only for QT? */
for (i = 0; lang[0] && !mp4 && i < FF_ARRAY_ELEMS(mov_mdhd_language_map); i++) {
if (!strcmp(lang, mov_mdhd_language_map[i]))
return i;
}
/* XXX:can we do that in mov too? */
if (!mp4)
return -1;
/* handle undefined as such */
if (lang[0] == '\0')
lang = "und";
/* 5 bits ASCII */
for (i = 0; i < 3; i++) {
uint8_t c = lang[i];
c -= 0x60;
if (c > 0x1f)
return -1;
code <<= 5;
code |= c;
}
return code;
}
They talk of "und" inside the code, but if you use -metadata:s:a:0 language="und" it just return English.. :\
What is the correct way to specify it? It could be a ffmpeg bug? I'm using the last zeranoe nightly build N-92405-ge24a754916
und
so it's not accepted. Will patch it soon.