## Are BRSTM files lossless?

1

Definition: .brstm is an audio file type that is commonly used on the Nintendo Wii and GameCube videogame platforms.

Question: Is BRSTM a lossy or lossless file type?

Although I've tried finding out through my own research, I haven't been able to determine much on this question:

• Although Wikipedia has no article on the subject, it mentions the term on the page, List of File Formats under the Lossless Audio header.

• However, the reference it cites, What Is a BRSTM File?, makes no mention of it being lossy or lossless.

• Additionally, the BRSTM technical page on the WiiBrew wiki, among the references of the above-mentioned reference, also makes no mention of it being either lossy or lossless.

1

From the line in WiiBrew 0x0008 0x20 16 Int16 ADPCM coefficients - ADPCM is a bit-reducing algorithm - https://en.wikipedia.org/wiki/Adaptive_differential_pulse-code_modulation though the math is way over my head, so I cannot be certain whether that is lossless or lossy - it might give you another avenue to investigate.

– Tetsujin – 2017-01-31T08:13:59.400

Wayyyyy over my head, too. Lol. Thanks, though. Maybe someone else with more knowledge than me can use this to figure it out, however. Haha. – SarahofGaia – 2017-01-31T12:50:30.547

1

The Binary Revolution Stream format uses ADPCM to represent audio data. ADPCM uses a lossless compression algorithm.

Differential pulse-code modulation (DPCM), like PCM, uses a cyclic pulse to sample a given waveform at discrete intervals. But instead of simply quantizing and storing the sample values, it takes the difference of two consecutive quantized samples and uses entropy coding to predict the next. The next sample value is then represented as a quantization of the difference from the predicted value (the prediction error value/entropy value). Therefore, this type of encoding reduces the bitrate, the more a pattern is observed. This means that no information lost.

In information theory an entropy encoding is a lossless data compression scheme that is independent of the specific characteristics of the medium.

Adaptive DPCM increases efficiency by using an adaptive scale factor to quantize the sample values.

So BRSTM is a lossless format; the trade-off is a smaller bitstream and file size for a greater amount of processing needed during encoding/decoding, which is a general trade-off for any lossless compression method.

Beautiful summary of using predictive coding in compression

1

From a quick read through several of the links above, I believe it's most appropriate to call BRSTM a lossless audio format. For a start, let's define "lossless audio":

For simplicity's sake, a lossless audio format is an audio format where the bits going in (during recording) match the bits going out (during playback).

To explain ADPCM (and I'm by no means an expert, so please someone correct me if this is incorrect), it scales the quantization levels, reducing the number of quantization levels necessary. Think of this as kind of like reducing bit depth (in an audio format like WAV), but only where it doesn't matter.

However, this algorithm is done at the input and output. So in BRSTM, ultimately, the bits going in are the same as the bits going out, and therefore it's a lossless audio format.

As an example, let's say I came up with a modified WAV audio format where the bit depth was reduced during quiet sections of the audio. This happens during encoding and during decoding, no lossy compression in between. So if you are only considering the format by itself, it's lossless. However, if you converted from another format (say "regular" WAV) then it would be lossy because bits would be dropped during quiets parts.

To sum up, it appears that BRSTM is best called lossless, but worth pointing out that converting from just about any other audio format would be a lossy process.

Hope that helps.

1Good effort, but your reasonings and explanations are wrong in a few places. e.g. Converting to another format would be just the same as converting from PCM. The only difference would be the conversion time. Also, "Think of this as kind of like reducing bit depth (in an audio format like WAV), but only where it doesn't matter" is wrong too. I won't down-vote, because it is, as you say, a lossless format. – Marc W – 2017-02-02T20:17:34.123