DSD v/s MQA – A new codec fight

DSD v/s MQA – A new codec fight

In the pursuit of improving the quality of audio delivered to the end user, there are two relatively new formats in the market today – Direct Stream Digital (DSD) and Master Quality Authenticated (MQA). Technically, they are very different, yet due to their advent in the same time frame, they are being compared inevitably.

Technology

DSD was primarily used on Super Audio CD media produced by Sony and Philips but now is distributed in DSF file format. It uses ultra-high rate of sampling with only single-bit wide samples using Delta-Sigma Modulation technique. This makes sure that the noisiest of analog signals are digitized faithfully. DSD uses Pulse Density Modulation encoding as opposed to Pulse Code Modulation, resulting in much higher fidelity. The sampling frequencies used are 2.8224MHz (DSD64), 5.6448MHz (DSD128), 11.2MHz (DSD256) and 22.5792MHz (DSD512). The audio signal can be recreated by simple long-term averaging of the input bitstream.

Due to the inherent nature of sigma-delta converters, the DSD and PCM streams cannot be compared directly. Yet for the sake of comparison, a DSD64 stream is approximately equivalent to a 20bit 96kHz PCM.

MQA is a lossy-compression based proprietary audio codec intended to render high fidelity digital audio. It can be contained in any lossless file format (eg: FLAC, ALAC), ensuring compatibility with non-MQA decoders. It encodes studio recorded PCM with authentication from the producer about the sound quality. Specialized ADCs reduce the quantization noises artifacts based on the dynamics of the sound and pass this information to the decoder.

The MQA encoder works on a unique principle of ‘audio folding’. The frequency spectrum is divided into three parts. First part represents frequencies up to 44.1kHz. The region above this frequency is parted into two bands (the scope of which is beyond this blog), each of which is folded and embedded within the noise threshold of the first part. The non-MQA decoders can decode the audio in the first part (only CD quality). The software-only decoders unfold the second and third parts and improve the audio quality. However, the best quality is presented by the complete MQA decoders using specialized DACs to replicate the dynamics of studio recording. The audio samples are compressed using some ADPCM based lossy encoding algorithm.

Hardware Requirements

Both DSD and MQA require specialized hardware. However, the requirement of special ADC and DAC is more stringent for DSD, as the sampling itself is changed. MQA can work with special ADCs improving the quality of captured sound. However, any normally recorded high resolution PCM data can also be encoded as MQA. Similarly, at the decoder, it is not compulsory to have special DAC to render the audio, but an MQA-DAC does help reproduce the exact sound as recorded in the studio with an MQA-ADC.

License

DSD is just a sampling method. DSF file format is an open specification. Any device with DSD ADC/DAC hardware can work with this kind of data. However, software conversion of DSD to PCM can enable any normal audio player to play this data with obvious loss of quality.

MQA is a proprietary closed-source codec specification licensed for encode/decode. While MQA codec is being appreciated by many for being remarkable in finding the balance between compression efficiency and audio quality, its stricter licensing term is being criticized across the industry.

– Nishit Jain

Privacy Preference Center