2000 Conference Proceedings

Go to previous article 
Go to next article 
Return to 2000 Table of Contents


Production Of DAISY Titles Using A/D Conversion

Markus Gylling
Lars Sönnebo
Swedish Library of Talking Books and Braille (TPB)



There is a need to start converting our archives from analogue to a digital format, mainly for two reasons. The first is that many of our analogue archives are deteriorating and we have to save our collections for the future. Many libraries and institutions serving print disabled people have started this transition using different solutions. The second reason is that, besides new production of digital talking books, analogue to digital conversion is the fastest and most important way to build up a large collection that you can offer the end users. This work also is going on with different solutions in different countries. This paper describes one way of A/D conversion used in Sweden by the Swedish Library of Talking Books and Braille.


| Top|

DAISY at TPB

Background

The everyday work for many years with talking book productions for e.g. university students is the prime mover behind the development of a new digital talking book system at the Swedish Library.

All the problems connected with studies using analogue talking books must be obvious to anyone working in this field. Today's analogue system is not suited for the professional use of recorded books, but as there is no other system it has been the technology used for many years.

Books in braille can not be red by everybody, and e-text is an option for very few.


| Top |

The Swedish Library

The Swedish Library of Talking Books and Braille (TPB) collaborates with local libraries to give people with print handicap access to literature. TPB is entirely government funded. The object of TPB is to produce and lend talking books and books in braille. Furthermore TPB provides advice and information on matters concerning talking books and braille.


| Top |

Service to college and university students in Sweden

TPB has a special service directed towards print handicapped students at a university level. This allows the students to have access to their course literature in the form of talking books, e-text books, braille, or enlarged text. These books, produced by TPB, can also be borrowed by professionals who wish to use them in their work.

Number of students 1998
Total amount of students

343

Blind and visually impaired

110

Mobility impaired

10

Dyslectic

223

Production of media 1998
New titles in total

1131

New analogue talking books

732

New braille titles

56

New e-text titles

249

New enlarged titles

94

 


| Top |

Improving the service to the students

At a certain point in the work with DAISY you have to go from project to reality and start producing large quantities of titles e.g. for the students.

Our library has decided to supply every university student with a digital talking book player, as a complement to the free reading software, Playback 2000, which students use when reading digital talking books with a computer. This will have the result that in a very short time nearly all of our approximately 400 university students will ask for digital DAISY books instead of ordinary cassette tapes.

To meet this demand TPB has decided to develop and set up an A/D unit called Tape Input Production System, TIPS. This system will have the capacity of around 1500 conversions in one year in a format we call raw-daisy. This means titles which are digitised and structured without any manual interaction in the production process.

The first goal for us is to deliver copies from the existing analogue archive in a DAISY format, which means around 1000 titles per year. The second goal is to produce a DAISY copy instead of lending the book from the shelf, which means an additional 4300 titles per year.


| Top |  

TIPS project at TPB 1999

In May 1999 we started to specify and plan the project. We estimated to have a unit running in the library from September the same year. One full-time operator was hired to handle the unit.

The following is a description of the equipment and components which we initially planned to use. Changes that were made at later stage of the project will also be described briefly.  

DAISY Tape Input Production System (TIPS)

Display of Tape Input Production System (TIPS) TPB version layout

 

 

Equipment list / component specification


| Top |

General

This system specification is for 8 channels of audio, using 4 master tape players, working in normal or double tape speed. Each tape player reproduces both tape sides simultaneously tape side ’A’ and tape side ’B’. Tape side ’B’ is played back in reverse order.

The design of this transfer system is based on the fact that TPB:s archive master tapes have been recorded on both sides. The full efficiency of this kind of system is dependent on this fact. If master tapes are only recorded on one tape side (1 track), the overall efficiency will be lower.

The system as such is not fixed to a certain master tape format, however. Any format can be used, as long as there are tape playback devices available for them. TPB uses ¼" tape in 2-track format.

The concept of doubling the transfer efficiency by playing back the archive tapes at double the standard tape speed, is dependent on the tape players offering this tape speed. In the case of TPB, the master tapes are recorded at 9.5 cm/s (3.75 ips), which means double speed playback is taking place at 19 cm/s (7.5 ips). Both these tape speeds are supported by most available open reel tape playback devices.

For the "double speed" transfer mode to work properly, it is important that the tape reproducers offers a full audio bandwidth, ideally with a high frequency response close to 20 KHz. When the tapes are played back at twice their normal speed, the frequency range of the audio will be shifted to double the normal frequency range. The high frequency response of the material when played back at normal speed (sample rate) will never be higher than about 10 KHz, but the frequency range might be a lot less than 10 KHz if the tape player can not transfer the highest frequencies to the A/D stage. As for the rest of the audio chain, all components are quite capable of handling frequencies up to 20 KHz, which means there will be no degradation of frequency range there, even when working at double tape speed.

The complete transfer station can produce 12-16X transfer rate. The actual figure is dependent on how efficiently the operator is able to mount new tapes and start new transfer sessions. This means that up to 16 playback hours of talking book data can be converted and processed in 1 working hour.

The complete transfer station can be operated by a single person. The same operator will be able to both control and monitor the actual transfer process, as well as being able to perform basic DAISY editing tasks. The most important editing operation will be to remove audio material that is directly related to the tape original, such as messages to change tape or flip tape sides.

The operator is also able to handle the administrative and practical tasks involved, such as backing up original data and writing distribution master CD:s.

The total efficiency of the system, as well as the information quality of the produced DAISY material, has proven to be highly dependent on the skills and efficiency of the operator. An operator must be able to edit the data fast enough to keep up with the transfer speed, or otherwise the total transfer speed will be lower than 12-16 X.

Of course, if there is a need for high information quality – that is, very well-edited and structured material – extra time can be spent on certain productions. Also, it is quite possible to use more staff in the production. Another person has been temporarily involved in editing dedicated to NCC structure.

An estimation of average production capacity is about 70 playing hours per working day. Since audio data compression is performed unassisted overnight, the station is able to deliver about 70 hours of output data in both uncompressed (though DAISY-structured) data for archiving as well as compressed distribution masters per day, provided that the equipment runs 24 hours per day.

This list mainly describes the audio equipment for the transfer station purchased by TPB, and is only intended as a general hint on the type of equipment to choose. Anyone who trusts in our competence is of course free to copy the set-up we are going to use. Others, who might have different (or even better) ideas can pick and choose in the list to build up their own system.


| Top |  

Audio equipment

Master tape players

High quality open reel tape playback devices that reproduces the archive tape in the format used by the particular organisation. As for TPB, 2-track, NAB format, 9,5 cm/s (3.75 ips) normal tape speed is used. Track 1 is here used for tape side A, while track 2 holds tape side B in reverse playback order. When working at double tape speed, 19 cm/s (7.5 ips) will be used.

TPB has several different machines available: Studer 902 , Studer B67 and Revox B77. At least one extra player will be included in the system, functioning as a spare unit. The total number of tape reproducers will therefore be 5 per transfer station.


| Top |  

Dynamic processors

2 Dbx 1046 compressors/limiters together offer 8 independent audio channels of dynamic processing. The device can be used both as a compressor/limiter and a peak limiting device simultaneously. The main purpose of the unit is to eliminate the risk of overloading the input of the mixer and hence the audio card of the PC: The units may also be used to reduce the dynamic range of some source materials, thereby offering a "tighter" sound.

The Dbx unit offers very high specifications for the analogue signal processing, such as great headroom and a very "musical" response of the compression/limiting circuitry. The intention is to set up the devices so that they will work well with virtually any type of archive material. The operator will then seldom or never have to change any settings on the devices.


| Top |  

Signal levels / connection type

As for the Studer tape players, balanced +4 dBu line level signals are used. The circuitry of the compresor/limiters and the inputs of the mixing console can work well with these types of signals. As for the Revox B77, it uses unblanaced –10 dBV line level signals. Both compressor/limiter and mixer can work equally well with these signal types, but have to be calibrated somewhat differently.


| Top |  

Cabling

XLR connectors and balanced cabling of high quality, preferably with gold-plated connectors. Horizon is one possible brand to go for, though there are many others. TPB has mainly bought cables from Horizon.

Some tape players, as e.g. the Revox B77, uses unbalanced connections. To connect these, special cabling will be needed (unbalanced RCA to XLR). High quality cables and connectors will help to avoid hum and other disturbances.

For a tidy set-up, it can be a good idea to get cables configured as multi-cables with 8 XLR leads in a collective cable.

The length of the cabling is not critical for balanced connections. It is a good idea to place the tape players and the compressor/limiter some distance away from the digital equipment to minimise the risk for disturbances.

As for unbalanced connections, it’s best to keep cable lengths down as much as possible. Keeping the analogue equipment away from computers etc. is still an issue, however.


| Top |

Mixing console

Yamaha 01V digital mixer, equipped with MY8-AT optical ADAT I/O expansion board. Optical fibre cable used for connection with the audio I/O card in the main DAISY processing PC.

The 01V offers 24 input channels, 8 busses and 4 output channels. The bus outs can be used as digital direct outputs using the expansion board. The mixer has high quality 20 bit A/D conversion and also has pre-amplification circuitry of high quality. Furthermore, the mixer is fully programmable and offers 4 bands of digital EQ as well as multi-type dynamic processing for every input and output channel.

The mixer is not used for mixing as such, but as a central A/D conversion and signal routing device. The inputs to the mixer come from the 8 audio channels of the master tape players and from the live recording microphone. The output goes to the multi channel (/O) board of the main processing PC and the sound card of the editing/control PC. Output also goes to reference monitor speakers and headphones.


| Top |

Multi-channel Digital audio I/O card

RME Digi 32/8 digital audio I/O card with Windows 95/98 and NT 4.0 drivers. Connected to 01V using optical ADAT interface and fibre cable.

The card can be used to feed up to 8 mono channels of 16 bit PCM audio at 44.1 or 48 KHz sampling rate from the mixer to the main processing PC. The PC accesses the card via its software driver as 4 stereo WAVE IN devices.

Optical fibre cable for the connection between mixer and audio card is included with the board, though it is not a very long one. If a longer cable is needed, they can be purchased separately. Be sure to get high quality optical fibre with appropriate connnectors for ADAT-interface device interconnection.


| Top |

Editing/live recording sound card

PC multimedia sound board, with S/P DIF digital connection to the mixer. Any high quality stereo sound card with digital S/P DIF input can be used for this, communicating digitally with the mixer at several different sampling frequencies. TPB happens to use the Gina card , produced by Echo and sold by Event Electronics. This card has 2 analogue inputs and 8 analogue outputs in addition to the 2 in/2 out digital connectors, which may be wasteful for the purposes of this system. However, we have found that the sound board seems to work fine within the set-up.

Several alternatives to the Gina card exist on the market. Some of these cards have only S/P DIF in and out, which may mean thy are less expensive and/or better suited for digital I/O.

The card should have driver software for windows 95/98, and preferably also for Windows N¤ 4.0. However, it is likely that the DAISY editing/control software will run under Windows 98.

Even though almost any sound card can be used, TPB can not recommend the SB live card, since its implementation of S/p DIF is limited. It is important that the card supports the sample rate(s) of interest for live recording, e.g. 22 KHz. This sample rate is actually not defined in S/P DIF, though on the Gina card, it works anyway. This is probably because 22050 is half the rate of 44100, which is the sample rate at which the mixing console is set up to work with as a "Wordclock master".

As an alternative to using S/P DIF input/output, normal analogue audio I/O may be used. The Yamaha 01V mixer offers a great number of analogue I/O connectors, which can be used to hook the mixer up to a purely analogue audio card of your taste.


| Top |

Reference monitor speakers

High quality reference monitor speakers are used for quality judgement and other reference listening tasks. We have chosen to use the powered monitor speaker systems from Event Electronics. These offer high audio quality and are convenient to use, since they have built-in power amplifiers. They can therefore be directly connected to the mixing console’s main stereo out or monitor out terminal. The speakers accept balanced XLR connectors.

We have got both the Event 20/20p and the Event Tria models at TPB, and can also recommend the 20/20bas model.


| Top |

Reference headphones

AKG K240 headphones are used for monitoring and reference listening e.g. to judge audio quality etc. They can be driven directly from the mixing console.

Microphone for live recording of extra messages etc.

We have chosen to use an AKG C 3000 microphone, directly plugged into the mixing console. Such a set-up does not offer full protection against overloading of the mixer’s input stage, since there is no analogue peak limiter in the audio chain. However, the 20 bit A/D converter in the mixer should offer enough headroom for most purposes, and the internal dynamic processors of the mixer can also be used to process the microphone signal.

Even though the microphone is of reasonably high quality, we have not gone for the ultimate device for this set-up. This is to save cost, but also keeping in mind that the main part of the recording work using this system will be for recording of short add-on material, such as for announcements and comments.

If highest possible recording quality is the goal, a more ambitious set-up may consist of a high-end microphone connected to a preamplifier with built-in compressor/limiter device and possibly also other analogue signal processors. The signal can then be connected tot an input of the mixer at line level.


| Top |

Software

The heart of the digital part of the system is a specialised version of the lpStudio software with an add-on software module for A/D operations.

The software performs many tasks in the system, including:

In addition to the specialised DAISY software, various commercial software packages for audio editing and processing was included in the setup, e.g. Sound Forge, Wavelab etc.

Audio analysis software was used to check frequency spectrum, dynamic characteristics etc.


| Top |

Hardware

Main DAISY processing PC. Dual 500 MHz Pentium Ii or Pentium III, running NT 4.0 for multi-processor use. 2x 16 Gb SCSI2 drives and 1x 5Gb Software/OS drive.

Editing / control / live recording PC: Pentium II 450, running Windows NT 4.0.

Backup / CD production PC: An ordinary Pentium II PC may be used, as long as it is fast enough to handle the task of efficient CD production etc.

CD-RW device/devices: 4X SCSI-2 units with appropriate CD mastering software.

CD-R device: Rimage Protege, a fully automated, two recorder CD-R handling system, including a thermoform Perfect print unit.

Tape backup device. At a production of 70 playback hours per working day, using 22050 Hz as sampling rate and using 16 bit PCM as sample format, the system gives a total output of about 11 GB of data per day. The tape device used for backup must therefore have enough capacity to safely hold large amounts of data, preferable about 10-20 GB per tape cartridge. DLT was used in our setup.

LAN: 100 Mbps Ethernet, suitable adapter cards, cabling etc. All PC.s in the system are connected to the network for data transfer. The network is not connected to other computers that are not directly involved with the DAISY production.


| Top |

Benefits and drawbacks of high automation transfer routines

This section reports on statistics and experience from 5 months (November 1999– March 2000) of full-time DAISY production with the TIPS system. Issues include


| Top |

Working with revisions of DAISY books

This section describes in-depth the workflow and practices of revision work, i.e. the subsequent refinement of NCC structure in already existing titles. Issues include a description of the use of LpStudioPlus and other software for these purposes, suggestions for quality and speed improvements (scanning, automated spellchecking etc), continuous validation, infrastructure (LAN and WAN transmissions of edited files), rebuilding on rewritable and WORM media, etc.


Go to previous article 
Go to next article 
Return to 2000 Table of Contents 
Return to Table of Proceedings


Reprinted with author(s) permission. Author(s) retain copyright.