How does LIBLTC support 16bit/24bit data encoding and decoding?

[MrWeng]

First of all, I am very grateful for providing an LTC encoding and decoding method. After my test, LIBLTC can only: encode and decode 8bit data, and does not support 16bit/24bit encoding and decoding. My steps are as follows:

1. Use the website LTC gererate to generate a 1-minute LTC audio file (Frame Rate: 23.976, Sample Rat: 480000, Bit depth: 16bit).
2. Use a binary editing tool to remove the header of the wav format, keep only the audio data (I think this is the source data of LTC), and modify the file suffix to raw.
3. Run ltcdecoder to decode the file, and only one message(* reading fromD:\libltc\test\LTC_02030405_1mins_23976fps_480000x16.raw) will appear after running,without any other information.
4. When I re-use the above website to generate an LTC audio file with bit depth=8bit, repeat steps 2 and 3. The LTC information can be parsed normally.The output information is as follows

• *reading from: D:\libltc\test\LTC_02030405_1mins_23976fps_48000x8.raw
  2000-00-00 +0000 02:03:04:05 | -23 2002
  2000-00-00 +0000 02:03:04:06 | 2003 4004
  2000-00-00 +0000 02:03:04:07 | 4005 6006
  2000-00-00 +0000 02:03:04:08 | 6007 8008
  2000-00-00 +0000 02:03:04:09 | 8009 10010
  2000-00-00 +0000 02:03:04:10 | 10011 12009
  2000-00-00 +0000 02:03:04:11 | 12010 14011
  2000-00-00 +0000 02:03:04:12 | 14012 16013
  2000-00-00 +0000 02:03:04:13 | 16014 18015
  2000-00-00 +0000 02:03:04:14 | 18016 20017
  I have two questions

1. Is the operation of the second step correct? Is the remaining data after removing the wav header file information the LTC data?
2. What should be done to enable libltc to realize the encoding and decoding of 16bit/24bit data?

[x42]

wav can contain multiple chunks, meta-data, and different encodings (signed, unsigned).
I highly recommend to use https://github.com/libsndfile/libsndfile (or a similar library) to get the raw PCM data from a wav file.

[MrWeng]

Thanks a lot, I've now been able to get and use PCM metadata correctly. But the sample format is 8bit unsigned mono, which cannot meet my needs. I want to change the sample format to 24bit mono, so what should I do?

[x42]

If it is signed 24 bit, divide each sample value by 256 and then use ltc_decoder_write_s16 for unsigned 24 bit you can simply shift each sample by 8 and use ltc_decoder_write_u16.

[MrWeng]

It may be that I didn't express clearly, I only use encoding, not decoding. The current encoded data is implemented based on 8bit unsigned data. If it is encoded based on signed 24bit data, how should it be handled? By the way, I did not understand the meaning of enc_lo and enc_hi in the structure LTCEncoder? What role do enc_lo and enc_hi play in the entire encoding process?

[x42]

For output you can likewise scale each sample:

ltcsnd_sample_t *buf;
ltc_encoder_encode_frame(encoder);
int len = ltc_encoder_get_bufferptr(encoder, &buf, 1);

for (int i = 0; i < len; ++i) {
  out_24bit_unsigned[i + off] = buf[i] << 16; // or ..
  out_24bit_signed[i +off] = buf[i] * 65535 - 8388480; //  * (2^16 -1) - 2^23
}

What role do enc_lo and enc_hi play in the entire encoding process?

LTC is a square wave. lo/hi are the signal levels for the waveform, you can change the default via ltc_encoder_set_volume()

Hope that helps.