In This Section
- Eastern Region, USA/Canada
- VP: Robert Breen
- Central Region, USA/Canada
- VP: Michael Fleming
- Western Region, USA/Canada
- VP: Jonathan Novick
- Northern Region, Europe
- VP: Bill Foster
- Central Region, Europe
- VP: Nadja Wallaszkovits
- Southern Region, Europe
- VP: Umberto Zanghieri
- Latin American Region
- VP: Valeria Palomino
- International Region
- VP: Toru Kamekawa
AES Section Meeting Reports
New York - October 15, 2013
With the recent release of 11.2 MHz Quad-DSD production tools, more than a decade of DSD and DXD productions and the rapidly growing availability of DSD and DXD material available for download on the market, there is a constant debate in both the professional and the audiophile sectors about the difference between DSD, DXD and PCM, and ultimately, which one "sounds better." Tonight's meeting, attended by about 25 AES members and guests, served as a preview of a full-length session on this topic to be held a few days later at the 135th AES Convention.
Dominique Brulhart of Merging Technologies introduced the discussion by describing his company's equipment which was used to make the first 11.2 MHz Quad DSD recordings in Boston 14 years ago. This work led to the creation of the SACD format by Sony. MT then had to create workstations to produce content. SACD has not survived as a general consumer delivery format due to the newer file-based formats such as MP3. It has, however, retained its audiophile user base. Current hardware is 5.6 MHz Double-DSD or the newer Quad DSD 256 at 11.2 MHz. Additional development work has yielded recordings in California in the DXD format using 352.8 KHz PCM 24-bit technology. This frequency is 8 times 44.1 KHz.
Next, John Newton of Soundmirror discussed the premise that the aim of recording technology, from Edison through vinyl, was to have the best quality original materials so that as technology improved this source material could be re-released to provide an ever-improving listening experience for the consumer. At this time both the producer and consumer can hear identical sound quality. In attempting to compare the relative merits of DSD, DXD and PCM formats the emphasis is now on how production workflow relates to the type of music being recorded. Format differences are now too small to be discernible in typical listening environments.
Editing accuracy is now 10 or even 100 times better than with previous technologies, thanks to a range of formats. Selection of technology firstly depends on the music, which determines the number of mics and individual tracks to be used. A simple Haydn symphony might only require the 12 MHz format. The recording equipment uses the Ravenna protocol which allows CAT-5E cable — a much cheaper method of shipping many channels of HD audio than traditional multicore copper cable. This cable is often left at the venue for future recordings. Two or three computers can easily be fed from the converters to simultaneously record backup originals. The protocol manages the sync function so unless there is a video production shoot at the session there is no need for external wordclock.
Morten Lindberg of 2L (Lindberg Lyd AS) agreed that the format decision depends on the workflow advantages of each. The converter is now working on raw sampling, so the focus is on the musicians and microphones. He works in DXD at live venues vs recording studios. His main mic arrays are usually the only mics needed. The most important tool is editing: there might be 2000 edits in one album project. Only about 10% of his 5.1 and 9.1 surround productions require equalization. A 2-track mixdown is created in post-production. Dynamics processing is realized by drawing manual envelopes onto the files. The preferred output format is pure audio Blu-ray discs. File size for a DXD 60-minute 2-track album is typically 2 GB. Albums are also delivered by download.
At the recording and editing phases there is no longer a traditional audio path — it's all data, and it can be handled without a computer. Patching is handled in the software. There is a need for increased security of the network on which this data is moving. This network can also include devices operating at various sampling rates. Additionally, latency has been reduced to the vicinity of 1.3 ms and is therefore "insignificant."
John Newton described his workflow: The original DSD recording is converted one time to DXD for editing and then reconverted one more time to DSD to create the SACD master. That is the limit of acceptable conversion. His company has 192 channels of DSD conversion in the field at any one time.
He added that some musical content, such as a cappella voice, is more revealing of differences in technology. DSD end-to-end takes 10 times longer for several reasons, one of which is that the equipment is still being developed and refined. Each edit is done in DXD and then converted back to DSD. This makes cost a major factor in both engineering development and production recording and editing phases of projects.
The second half of the evening was devoted to listening to samples of recordings made by Morton and John. We saw the sound files themselves as well as photos and diagrams from the sessions. Kinds of microphones were discussed (mainly B&K/DPA omnis), as were as the recording venues and how they were selected. Morton explained that Merging Technologies, recording in DXD at 352.8 KHz, 24 bits, had created its own record label and publishing company "2L" and has released 100 titles since 2002.
Additionally, we learned that there are six different types of church available for recording in the Oslo area. Each project finds the venue most suited to the music being recorded. In all cases the signal path was as follows:
mic > (custom) preamp > converter > recording computer.
We heard a playbacks of a small jazz combo with and without vocal; solo piano; and finally a military band — all recorded in the same church. All of the evening's playbacks were from a dedicated stereo mix in which the left and right front mics predominated, with a small amount of the center and surround mics added in. The listening experience for the 2L material is atypical in that the listener is surrounded by the ensemble, not in front of it. "This is a more emotional experience," he said.
John agreed. "After all, a recording is a different experience than a concert." Composers who have already passed away could not be expected to envision such a means of performing their music. Mort observed that when he works with living composers they often collaborate with him by re-writing certain passages to enhance the end-user listening experience by exploiting the possibilities of this new technology.
John's playbacks were from files recorded at 12 MHz in DSD. Only the crossfades (edits) were converted to DXD — perhaps 10ms in length each -- and reconverted to DSD. They were heard without processing as they are straight from just-made original recordings. We heard a Philip Glass piece for two trumpets and a Haydn symphony. These sessions will eventually be released by Pentatone as discs and downloads.
Audience members peppered the presenters with questions throughout the evening and were rewarded with detailed answers. A lively post-presentation discussion continued for at least 30 minutes.