AMI Digital Asset Technical Specifications

This document outlines the technical specifications and requirements for digitization of analog media collections and digital packaging of deliverable files. For corresponding information about metadata; shipping, handling and reporting; and project administration, please see:

• ami-handling
• ami-metadata

Table Of Contents

• Specifications for Digital Assets
  • Digital Asset Packaging
  • Metadata
  • Film Media
    • Film groups 1 & 2: motion picture film
      • Preservation masters: film groups 1 & 2
      • Mezzanines: film groups 1 & 2
      • Service copies: film groups 1 & 2
    • Film group 3: audiofilm
      • Preservation masters: film group 3
      • Edit masters: film group 3
    • Film group 4: filmstrip
  • Video Media
    • Video preservation masters
      • Video group 1: analog and digital cassette, analog reel
      • Video group 2: DV cassette
      • Video group 3: video optical disc
      • Service copies: all groups
  • Audio Media
    • Audio preservation masters
      • Audio group 1: analog magnetic
      • Audio group 2: digital magnetic
      • Audio group 3: audio optical disc
      • Audio group 4: grooved disc
      • Audio group 5: grooved cylinder
      • Audio edit masters: all groups
  • Data Media
    • Data group 1

Specifications for Digital Assets

Specifications may be modified over time to reflect changes in best practices or NYPL’s digital infrastructure, or to reflect previously unspecified media or conditions.

The following sections are broken into format groups to define the file deliverables for different media types and format variations.

Digital asset packaging

Package digital assets using the BagIt specification (V0.97) (https://tools.ietf.org/html/rfc8493).

Bag Requirements

• The Bag payload must contain the digital assets organized within a directory structure described by NYPL.
• The Bag directory must be named with the Primary ID of the collection object represented by the files in the bag.
• The Bag must contain an md5 manifest that lists every file in the data directory and their md5 checksums.
• The Bag must contain all of the media and metadata files from one, and only one, inventoried collection object.
• The Bag should be created with a tool using BagIt Library 4.4 or higher.
• The Bag may include a tagmanifest.txt file. Some bagging software produces this file automatically. It is not required to produce or * delete these files.
• The Bag and the files within it must not be compressed with zip, tar, etc.

Subdirectories

• Sub-directories must be exactly named as they are written here, in plural form, regardless of the number of files within the directory.
• Film Bags must contain the following subdirectories within the data directory:
  • PreservationMasters
  • Mezzanines
  • ServiceCopies
• Video Bags must contain the following subdirectories within the data directory:
  • PreservationMasters
  • ServiceCopies
  • Images
• Audio Bags must contain the following subdirectories within the data directory:
  • PreservationMasters
  • EditMasters
  • Images
• Data Bags must contain the following subdirectories within the data directory:
  • PreservationMasters
  • Images
  • Files
• For each preservation master file in the PreservationMasters directory, corresponding derivative file/s must exist.
• Each audio, film, video media, or data file in the Bag must have a corresponding JSON metadata file (image files do not require metadata).
• Example bag structure:

File naming convention

NYPL will provide the vendor with a filename “root” for each collection object, consisting of a three-letter prefix, the primary ID, a volume number (and a face number for audio objects), and a two-letter suffix indicating the role of the file. These sections are each separated by an underscore: (prefix)(primary ID)(components)_(file role).ext

Video example: myd_123456_v01_pm.mkv Audio example : myh_987654_v01f01r01_pm.flac

It is the vendor’s responsibility to complete the filename to accommodate multiple volumes, faces, regions, etc. where applicable, as defined by the filename components listed below.

File role

A media file name must record its intended role using one of the following suffix codes: pm: Preservation Master, created for every physical media object mz: Mezzanine, created for every motion picture film (film groups 1 and 2) em: Edit Master, created for audio files sc: Service Copy, created for video files image files do not require a role suffix

Metadata

NYPL Metadata inventory

NYPL will provide the vendor with a Microsoft Excel spreadsheet containing identifying and descriptive information about each NYPL collection object included in the PO. Note that one collection object with a single unique primary id may contain multiple physical objects (for example: a double compact cassette case).

NYPL JSON schema

NYPL metadata deliverables must adhere to customized fields and controlled vocabulary defined within NYPL’s JSON Schema. The schema, sample files, and validation instructions are hosted on GitHub: https://github.com/NYPL/ami-metadata. Metadata must be packaged as a single JSON file for each audio, video, or film media file (Images do not require JSON metadata). Metadata must validate against JSON schema. Any invalid files or errors will be corrected by the vendor and require redelivery of an entire Bagged digital asset.

• Metadata files must be named with the same root as the media file to which they pertain, but must not include the original extension of the media file. Example:
  • Correct: [filename].json
  • Incorrect: [filename].mov.json
• The vendor will be responsible for the following:
  • Maintaining a system that is able to easily sync with and incorporate updates from NYPL’s live GitHub repository, ami-metadata, and
  • Generating metadata about the source object, preservation master and derivative files, treatment and digitization processes, and the transfer operator.

Metadata content

• The required contents of the metadata files are defined by the JSON metadata schema. According to that schema, metadata will be produced for each deliverable that includes the following designated categories:

  • Bibliographic (provided by NYPL)
  • Source (partially provided by NYPL)
  • Technical
  • Digitization process
  • Digitizer

• Metadata must use the only the fields and values defined in the schema.

• If there is a question about a field, if the provided vocabulary does not accurately describe an object or process, or if a list of terms is insufficient, contact NYPL for guidance on how to proceed. NYPL may approve and release new terms on the schema depending on priority.

• Metadata fields must not be empty. Refer to the schema or contact NYPL for allowances of “unknown” as a value.

Metadata notes fields

The metadata schema includes a number of ‘notes’ fields which must be used as described below:

• bibliographic.accessNotes: NYPL added. No additional input required
• bibliographic.contentNotes: any additional content notes (“title program ends mid-tape, followed by another program”; "contains explicit content")
• technical.signalNotes: audiovisual signal characteristics noticed during capture ("tracking errors; audio buzz")
• source.notes.physicalConditionDigitizationNotes: description of pre-existing damage or decay not already noted in PreShip notes (“broken leader”; "bad splices")
• source.notes.physicalConditionPreShipNotes: NYPL added. No additional input required
• digitizationProcess.notes.processNotes: rehousing, adding leader, baking, cleaning

Metadata errors

If the technical characteristics or format of an object have provided by NYPL are incorrect (i.e. metadata inventory describes the format as “video cassette analog” and it is actually “audio cassette digital” with a format of “Umatic/PCM”), please provide corrected metadata. Make any changes to source object metadata elements that would be appropriate, and note the correction in the "digitization.notes.processNotes". Example text: "NYPL-provided metadata incorrectly listed [insert field name here] as [insert wrong content here]; JSON reflects correct [field name]"

BEXT Metadata

During audio digitization, the following BEXT metadata must be captured in the original Broadcast Wave files to ensure retention in the final FLAC deliverable.

BEXT field	Explanation	Sample data
Description	Description of who created the file	File generated by Vendor Services, US
Originator	Country code, Institution	US, The New York Public Library
OriginatorReference	technicalFileName	myt_123456_v01f01_pm
OriginationDate	Date file created in YYYY-MM-DD format	2017-06-23
OriginationTime	Time file created in hh:mm:ss format	11:20:47
BextVersion	[version number]	1
CodingHistory	Signal chain from which the digital file was created, starting with the analog or digital source. There are six elements that can be included in the coding history:1. A = coding algorithm (analog, PCM, etc.); 2. F = sampling frequency in Hz; 3. B = bit rate (for MPEG only); 4. W = word length or bit depth; 5. M = mode or sound field (mono, stereo); 6. T = free text to describe playback and capture equipment	A=PCM,F=44100,W=16,M=stereo,T=TSSTcorp DVDRW SH-216BB; Compact Disc; A=PCM,F=44100,W=16,M=stereo,T=dBpoweramp 14.3; Import CDDA as WAV

Film Media

Deliverables

For each original recording, the following shall be produced:

• One preservation master file*
• One mezzanine file*
• One service copy file*
• One metadata file per media file If the object has audio content (i.e. composite sound print), audio must be synchronized and embedded in all final deliverables.

Capture tools

Film must be digitized and captured as DPX, then transcoded with any synchronous Broadcast Wave files to FFV1/FLAC/Matroska using RAWcooked (https://mediaarea.net/RAWcooked).

Condition Information NYPL will provide item-level condition information where possible in the following inventory fields. Note - only "condition_splices" has a scale that differs from the generic "minor > severe" scale:

• condition_splices:
  • 0: no splices
  • 1: 1 - 5 splices
  • 2: 6 - 10 splices
  • 3: More than 10 splices
• condition_perforation_damage
  • 0: non-existent
  • 1: minor
  • 2: moderate
  • 3: severe
• condition_distortion
  • 0: non-existent
  • 1: minor
  • 2: moderate
  • 3: severe
• condition_fading
  • 0: non-existent
  • 1: minor
  • 2: moderate
  • 3: severe
• condition_scratches
  • 0: non-existent
  • 1: minor
  • 2: moderate
  • 3: severe

Film Groups 1 and 2: Motion picture film, silent / sound

Preservation master file specifications: Film groups 1 & 2: Motion picture film, silent / sound

Source format	35mm	16mm	8mm / Super 8mm / Double 8mm
Video bit depth	16 bit	10 bit	10 bit
Resolution	4K optical side overscan	2K optical side overscan	2K optical side overscan
Video codec	FFv1 version 3 from DPX	FFv1 version 3 from DPX	FFv1 version 3 from DPX
File wrapper	Matroska (.mkv) from DPX	Matroska (.mkv) from DPX	Matroska (.mkv) from DPX
Frame size	4096x3112	2048x1556	2048x1556
Frame rate	(Same as source. If not described, please determine on viewing and describe in metadata signal)	(Same as source. If not described, please determine on viewing and describe in metadata signal)	(Same as source. If not described, please determine on viewing and describe in metadata signal)
Pixel aspect ratio	1.000	1.000	1.000
Audio bit depth*	24 bit	24 bit	24 bit
Audio sampling rate*	96,000 Hz	96,000 Hz	96,000 Hz
Audio codec/data encoding*	FLAC from PCM/WAV	FLAC from PCM/WAV	FLAC from PCM/WAV
Audio channels*	Same as source object	Same as source object	Same as source object
Color space	Linear RGB	Linear RGB	Linear RGB
Color primaries	BT.709	BT.709	BT.709
Transfer characteristics	Printing Density	Printing Density	Printing Density
Notes	Transcoded to FFV1/MKV from DPX using RAWCooked	Transcoded to FFV1/MKV from DPX using RAWCooked	Transcoded to FFV1/MKV from DPX using RAWCooked

Mezzanine file specifications: Film group 1 & 2 (Motion picture film, silent / sound)

Source format	35mm	16mm	8mm / Super 8mm / Double 8mm
Bit depth	10 bit	10 bit	10 bit
Resolution**	1920 x 1080	1920 x 1080	1920 x 1080
Display aspect ratio**	16:9 pillarboxed / letter boxed as needed	16:9 pillarboxed / letter boxed as needed	16:9 pillarboxed / letter boxed as needed
Video codec	ProResHQ	ProResHQ	ProResHQ
File wrapper	Quicktime	Quicktime	Quicktime
Frame size**	1920 x 1080	1920 x 1080	1920 x 1080
Frame rate	(Same as preservation master)	(Same as preservation master)	(Same as preservation master)
Scan type	Progressive	Progressive	Progressive
Pixel aspect ratio	1.000	1.000	1.000
Audio data encoding*	PCM	PCM	PCM
Audio bit rate*	2304 kbps	2304 kbps	2304 kbps
Audio bit depth*	24 bit	24 bit	24 bit
Audio sampling* rate*	48,000 Hz	48,000 Hz	48,000 Hz
Audio channels*	same as Preservation Master*	same as Preservation Master*	same as Preservation Master*
Color space	4:2:2	4:2:2	4:2:2
Image corrections	Color corrected for dye fading, cropped to picture - no frame-lines or sound track visible, Non-anamorphic	Color corrected for dye fading, cropped to picture - no frame-lines or sound track visible, Non-anamorphic	Color corrected for dye fading, cropped to picture - no frame-lines or sound track visible, Non-anamorphic

* Where audio is applicable. ** Subject to change. TBD.

Service copy file specifications: Film group 1 & 2 (Motion picture film, silent / sound)

Source format	35mm	16mm	8mm / Super 8mm / Double 8mm
Bit depth	8 bit	8 bit	8 bit
Resolution**	720 x 480	720 x 480	720 x 480
Display aspect ratio**	4:3 letterboxed as needed	4:3 letterboxed as needed	4:3 letterboxed as needed
Video codec	H264	H264	H264
File wrapper	MPEG-4 (.mp4)	MPEG-4 (.mp4)	MPEG-4 (.mp4)
Color space	4:2:0	4:2:0	4:2:0
Frame size**	720 x 486	720 x 486	720 x 486
Frame rate	(Same as preservation master)	(Same as preservation master)	(Same as preservation master)
Scan type	Progressive	Progressive	Progressive
Pixel aspect ratio	1.000	1.000	1.000
Audio codec*	AAC	AAC	AAC
Audio bit rate*	320 kbs	320 kbs	320 kbs
Audio sampling rate*	48,000 Hz	48,000 Hz	48,000 Hz
Audio channels*	same as Mezzanine*	same as Mezzanine*	same as Mezzanine*
Image corrections	Color corrected for dye fading, cropped to picture - no frame-lines or sound track visible, Non-anamorphic	Color corrected for dye fading, cropped to picture - no frame-lines or sound track visible, Non-anamorphic	Color corrected for dye fading, cropped to picture - no frame-lines or sound track visible, Non-anamorphic

* Where audio is applicable. ** Subject to change. TBD.

Film Group 3: Audio film

Preservation master file specifications: Film group 3 (Audio film)

Source format	35mm	16mm	8mm / Super 8mm / Double 8mm
Audio data encoding	Free Lossless Audio Encoding (FLAC)	Free Lossless Audio Encoding (FLAC)	Free Lossless Audio Encoding (FLAC)
Wrapper	FLAC (.flac)	FLAC (.flac)	FLAC (.flac)
Bit depth	24 bit	24 bit	24 bit
Sampling rate	96,000 Hz	96,000 Hz	96,000 Hz
Number of audio channels	(same as source)	(same as source)	(same as source)
Other characteristics	If there are tones / sync marks present, they must be captured or resolved and described in metadata signal notes.	If there are tones / sync marks present, they must be captured or resolved and described in metadata signal notes.	If there are tones / sync marks present, they must be captured or resolved and described in metadata signal notes.
BEXT metadata	If possible, BEXT metadata must be embedded as per specifications for audio deliverables	If possible, BEXT metadata must be embedded as per specifications for audio deliverables	If possible, BEXT metadata must be embedded as per specifications for audio deliverables

Edit master file specifications: Film group 3 (Audio film)

Source format	35mm	16mm	8mm / Super 8mm / Double 8mm
Audio data encoding	Flac	Flac	Flac
Wrapper	Flac	Flac	Flac
Bit depth	equal to preservation master	equal to preservation master	equal to preservation master
Sampling rate	equal to preservation master	equal to preservation master	equal to preservation master
Number of audio channels	equal to preservation master	equal to preservation master	equal to preservation master
Other characteristics	If there are tones / sync marks present, they must be captured and described in metadata signal notes	If there are tones / sync marks present, they must be captured and described in metadata signal notes	If there are tones / sync marks present, they must be captured and described in metadata signal notes
BEXT metadata	BEXT metadata must be embedded as per specifications for audio deliverables	BEXT metadata must be embedded as per specifications for audio deliverables	BEXT metadata must be embedded as per specifications for audio deliverables

Film Group 4: Filmstrips

NYPL will review recommendations for digitization of filmstrips (and accompanying audio media, where applicable) before defining a specification. Requests for recommendations regarding these objects have been included in the Business Questions section.

Video media

Deliverables

For each original recording, the following shall be produced:

• One preservation master file
  • If captions are present in source, one closed captions sidecar file
• One service copy file
• One metadata file per media file
• Image files as described

Capture tools Preservation master video files must be generated by professional-grade capture devices and software, with either direct capture to FFV1/FLAC/MKV, or transcoding from V210/PCM/MOV. Specific FFmpeg transcoding recipes will be provided by NYPL to ensure consistency.

Guidelines: Preservation master files, all groups

• Characteristics intrinsic to the broadcast standard of the source material, including frame rate, pixel aspect ratio, interlacing, resolution, and recording standard (NTSC, PAL, SECAM, etc.) should be preserved.
• Signal extraction must be optimal, and carried out using the equipment and accessories that are appropriate for the original format characteristics.
• The most direct and clean signal path must be used at all times from source to destination. There may be no devices inserted in the signal path that are not being used. If there are multiple destination formats being used in the transfer the signal path must be routed in parallel. No daisy-chaining of devices may occur.
• The highest quality signal format (composite, S-Video, Component, SDI, etc.) available for the source media type must be used throughout the entirety of the signal path from source through destination. Exceptions to this must be explained and requested prior to performing the transfer.
• Luminance, black, and color levels should be adjusted to existing color bars if they are present on tape and look accurate. If color bars are not present or are clearly inaccurate, preview each tape in order to adjust levels according to the content of the tape using known references (such as blue sky, known blacks and whites, flesh tone, etc.). Luma should be adjusted to fall within broadcast range (100 IRE max) and must not exceed 110 IRE.
• The transfer should capture all content recorded on the original object, including any bars and tone, slates, or other material coming before the start of the recorded program.
• The recording should run until the end of the recorded content (picture and sound). If this endpoint cannot be unambiguously determined, the recording should run until the end of the original object.
• If present on the source tape, closed captions must be captured.

Video preservation master sidecar files

Closed captions

• If captions are present in source object, an .scc sidecar file must be created and accompany the preservation master file (closed captioning must be embedded in service copies. See service copy specifications).
  • Format: Scenarist Closed Caption File (.scc)
  • Naming Convention: division_PrimaryID_v01_pm.scc
• Closed captions specifications may change in the future to accommodate updates to standards or NYPL infrastructure needs.

QCTools Reports

• Each video preservation master file should receive a corresponding QCTools report, which will be included in the PreservationMasters Bagged directory as a sidecar file.
  • Format: QCTools Report (gzipped XML) (https://www.bavc.org/preserve-media/preservation-tools)
  • Naming Convention: division_PrimaryID_v01_pm.mkv.qctools.xml.gz

Preservation master file specifications: video group 1: analog and digital cassettes, analog open reel

Attribute	Specification
Video codec	FFv1 version 3
Data compression	Lossless, Intra-frame (GOP-1) only
Chroma subsampling	4:2:2 YUV
Bit depth	10-bit
File wrapper	Matroska (.mkv)
Frame rate	(Same as original media)
Frame size	(Same as original media)
Broadcast standard	(Same as original media)
Pixel aspect ratio	D1 NTSC (.91) or PAL (1.09)
Slices	24
Slicecrc	1
Audio format	FLAC
Audio bit depth	24-bit
Audio sampling rate	48 kHz
Audio channels	(Same as original media, see guidelines for silent channels)

Silent audio channels

• If a channel exists and is silent, the channel should be captured and a note should be included in the JSON file that the indicates that the tape and the channels delivered on preservation master file are silent. If a channel is non-existent, then it won't be captured. For example:
• If soundField="ch.1:mono, ch.2: none" rather than soundField= "mono", then the second channel exists, and source.audioRecording.numberOfAudioChannels = 2 (and the preservation master file delivered will contain 2 channels of audio, one of which is silent).
• Two silent channels If detected as actual channels / i.e. recorded with "black" vs. not recorded), both channels should be captured and delivered with the preservation master, and a signalNote should be included that the tape is silent.

Timecode

• Two forms of legacy/source timecode should be retained: LTC Timecode, recorded on an audio channel, should be captured as an audio stream in the resulting preservation master file; VITC timecode, if present, should be captured through the use of appropriate playback devices and a carefully routed SDI signal chain.

Preservation master file specifications: video group 2: DV (digital video) cassettes

NYPL prefers native capture of DV content, with the understanding that errors and varying conditions may require alternative approaches to signal capture. Below are NYPL primary specifications, followed by specifications for cases where the native DV capture with .dv wrapper is problematic due to problems with a given object.

Attribute	Specification	Notes
Video codec	(Same as source)	See note
File wrapper	DV (.dv)	See note
Other characteristics	(Same as source)	If it is necessary to capture DV or HDV tapes via SDI (i.e. as V210/MOV), due to problematic tapes, deliverables must be transcoded and rewrapped as “.mkv”.

Preservation master file specifications: video group 3: optical discs

Attribute	Specification
File system	ISO 9660 / UDF
File wrapper	ISO (.iso)
Other characteristics	(Same as source)

Service copy file specifications: all video groups

Attribute	Specification
Video codec	H.264/MPEG-4 AVC (ISO/IEC 14496-10 - MPEG4 Part 10, Advanced Video Coding)
Video bit rate	3.5 Mbit/second
Chroma subsampling	4:2:0 YUV
Bit depth	8-bit
File wrapper	MP4
Frame rate	(Same as preservation master)
Frame size	(Same as preservation master)
Broadcast standard	(Same as original media)
Pixel aspect ratio	(Same as preservation master, see below for Anamorphic video)
Audio codec	AAC
Audio bit rate	320 kbps
Audio sampling rate	48 kHz
Audio channels	2 (see examples)
Closed captions*	CEA-608 (*if applicable)

Service copies, video group 3: optical video disc

• An individual service copy must be created for all discreet content. Example:
  • If a disc contains two discreet videos, each with different display aspect ratios (i.e. 16:9 vs. 4:3), a separate service copy must be made for each video.

Additional video specifications

Anamorphic video

• For service copies created from Anamorphic preservation masters, treat source as D1/DV NTSC or PAL Widescreen to produce a 16 x 9 service copy without padding. Pixel aspect ratio should be 1.21 (NTSC) / 1.46 (PAL).

Complex audio configurations

• 4 audio channels on source:
  • If a source tape has 4 audio channels of identical content (i.e. 4 mics in a single room recording the same content), the 4 channels captured in the preservation master should be mixed down to 2 for the service copy.
  • If each audio channel contains different content, consult with NYPL for how to proceed.
    • 2-channel Spanish / English language audio: If one channel contains Spanish dialogue and the other channel contains English dialogue, the audio content of the preservation master and the service copy must be identical.
  • Audible content in only one channel:
    • If there is only audible content in a single channel of any number of channels in a source tape and preservation master, the channel containing audible content should be mapped for production of a 2-channel service copy to provide a better user experience.

Timecode

• Audible LTC Timecode should be eliminated, and “normal” content should be duplicated onto the second channel.

Trimming of Heads and Tails

• Color bars must not be trimmed.
• Long tails of black / “snow” / unrecorded content may be trimmed after confirmation that there is no visible or audible recorded content.
• Trimming must not result in an abrupt end of visible or audible content.

Audio media

Deliberables

For each collection object, the following shall be produced:

• One or more preservation master file(s)
• One edit master file per preservation master
• One metadata file per media file
• For optical audio only, one CUE file per preservation master file
• Image files as described

Capture tools

Preservation master and edit master files must be captured/encoded as Broadcast Wave Format (BWF), with embedded BEXT metadata (see “Embedded Metadata” section).

Files that exceed the Broadcast Wave Format 4GB file size limitation should be captured as RF64, with the same embedded metadata values. Post-capture, all Broadcast Wave and RF64 files should be transcoded to the FLAC codec and container, with embedded metadata and original modification times retained through the use of the FLAC Utility (https://xiph.org/flac/download.html) by following the command listed on the FFmprovisr website (https://amiaopensource.github.io/ffmprovisr/#flac-tool). Original capture as Wave64 (.w64) is not acceptable.

Preservation master

• Technical guidelines: The production of preservation master files will comply with the technical recommendations, practices and strategies outlined by the International Association of Audiovisual Archives
• Strategic guidelines - IASA-TC 03, version 3: The production of preservation master files will comply with the ethical recommendations, practices and strategies outlined by the International Association of Audiovisual Archives.
• Optimal signal extraction from analog sources seeks to be complete, and includes the transfer of the “lead-in” and “play-out” portions of a recording.
• Analog signals will be converted to a digital bitstream by means of an Analog-to-Digital converter which complies with the specifications in FADGI’s Audio Analog-to-Digital Converter Performance Specification and Test Method.
• No signal processing will be applied to the Analog-to-Digital converter’s digital bitstream, including, but not limited to equalization, level adjustment, dither, noise reduction.

Signal Extraction

• Signal extraction from analog original audio recordings will comply with the technical recommendations, practices and strategies outlined by the International Association of Audiovisual Archives.
• Optimal signal extraction for the production of preservation master files should aim to capture the complete dynamic and frequency ranges of the original recording.
• Signal extraction must be carried out using the equipment and accessories that are appropriate and intended for the original format characteristics.
  • Example: a full-track mono recording on an open reel audio tape must be transferred using a full-track audio head (rather than a stereo head).
  • If this is not possible, provide an explicit proposal for work, subject to NYPL approval.
• An optimal signal extraction from the original recording will be a flat (unmodified) transfer, free of signal processing, equalization, level adjustment, noise reduction, etc. Exceptions would include:
  • de-emphasis of a recording’s stated pre-emphasis (playback equalization)
  • decoding of a recording’s stated noise-reduction encoding
  • Optimal signal extraction from original sources includes the extraction of the intended signal, along with any unintended signal (such as artifacts and anomalies in the signal associated with the inherent limitations of historic recording technologies).
  • Preservation master files may be one or two-channel (interleaved), and the configuration employed will be determined by the needs of the original recording.
  • Levels may be adjusted ONLY if there is severe distortion or digital clipping from the source, and this adjustment must be noted clearly in the metadata.
  • If a sync tone is present (i.e. Pilottone), tone must be captured or resolved.
  • For all grooved media, preservation master files must include the “needle-drop” and “needle-lift”.

Special circumstances and object-file relationships

• Faces: In general, one preservation master file will be generated for each physically or technically discrete recording area of the original object.
  • For example, each side of an audio cassette or disc will be recorded as a separate preservation master file, identified with a designated Face number.
• Considerations and circumstances that impact the number of preservation masters generated for each physical object, and/or for each discrete recording area:

  • For an object which is found to have content recorded both forwards and backwards on the same Face, a second Face may be created for the reversed content to be recorded in the proper direction.

  • Region: For an object with regions recorded at different speeds or sampling rates, a separate preservation master must be created for each Region (filename_v01f01r01_pm).

    • File overlap:
      • Speed Changes: If multiple preservation masters are created for a single recording due to speed changes, the cut should be made at a logical break in the audible content (if at all possible), and there must be exactly 5 seconds of audible content overlapping between the tail of the first preservation master and the head of the following PM so that the regions may be recombined in the future if necessary.
      • Sampling Rate Changes: If a digital source object has been recorded at multiple sampling rates, a separate preservation master must be created for each region. These regions do not need to overlap, but please include a note listing the timestamp on the source object where each region begins.

Embedded metadata

BEXT metadata must be embedded in each audio file upon capturing to WAV. See Metadata section for details.

Preservation master file specifications: audio group 1: analog magnetic

Attribute	Specification
Audio data encoding	Free Lossless Audio Codec (FLAC)
File wrapper	FLAC (.flac)
Bit depth	24
Sampling rate	96,000 Hz
Number of audio channels	(same as source)

Preservation master file specifications: audio group 2: digital magnetic

Attribute	Specification
Audio codec	Free Lossless Audio Codec
File wrapper	FLAC (.flac)
Bit depth	(same as source)
Sampling rate	(same as source)
Number of audio channels	(same as source)

Preservation master file specifications: audio group 3: optical disc

Attribute	Specification
Audio codec	Free Lossless Audio Codec (FLAC)
File wrapper	FLAC (.flac)
Bit depth	(same as source)
Sampling rate	(same as source)
Number of audio channels	2 (left + right discrete)
Other characteristics	CDs should be captured as a single file

CUE sheet files

In congress with capturing a Broadcast Wave file prior to transcoding to .flac, a CUE file must be generated. The CUE file must:

• Follow the same naming convention as the WAV file, but instead with a ".cue" extension. Example: "myh_123456_v01f01_pm.cue"

• Be referenced in the JSON file under the technical.cueFile field, by its complete filename

• Be nested within the Preservation Masters directory, accompanying the Preservation Master WAV file (the Edit master must not have a .cue file):

  • PrimaryID
    • data
      • PreservationMasters
        • division_PrimaryID_v01f01.flac
        • division_PrimaryID_v01f01.cue
        • division_PrimaryID_v01f02.flac
        • division_PrimaryID_v01f02.cue
      • EditMasters
        • division_PrimaryID_v01f01.flac
        • division_PrimaryID_v01f02.flac

Preservation master file specifications: audio group 4: grooved disc

Attribute	Specification
Audio data encoding	Free Lossless Audio Codec (FLAC)
File wrapper	FLAC (.flac)
Bit depth	24
Sampling rate	96,000 Hz
Number of audio channels	2 (left + right discrete)

Reproduction details

• Playback EQ curves for all discs should be utilized.
  • If a disc’s playback curve is not known or stated on the label, the phono EQ should be set to RIAA (for microgrooves 1948-present) or a “default” curve of 400Hz turnover and -12dB @ 10kHz rolloff (for transcription discs). All EQ curves should be noted in the metadata digitization process notes.
• Equipment
  • NYPL requires either the Time-Step line of phono EQs, or the KAB Souvenir Equalizer. Owl EQs are not acceptable due to the age, added noise floor, and lack of versatility needed for our collections. Other phono EQs may be submitted for approval.
• All discs should be transferred with the cartridge switched to "lateral" unless otherwise indicated on the label.
• A full complement of stylus shapes and sizes should be on-hand at the vendor transfer station, and he best sounding size/shape stylus used for transfer should be noted in the metadata digitization process notes. * Modern LP styli must not be used to transfer shellac, shellac-vinyl compound, or early vinyl discs.
• Tone arm requirements: * Arm must be long enough to comfortably reach the outer grooves. * Arm should be vertically adjustable to accommodate the various thicknesses of pressings. * During playback, the angle of the tonearm from front to back should be perfectly parallel in relationship to the disc while it is playing.
• While playing a face, for any speed changes or instances where the stylus has to be repositioned to continue playing, a separate preservation master must be created for each Region.
  • If the stylus skips, attempt to correct and retransfer the area before and after, add this note into the metadata along with time stamp.
    • If unable to perform these functions, set aside and return to NYPL.

Preservation master file specifications: audio group 5: grooved cylinder

Attribute	Specification
Audio data encoding	Free Lossless Audio Codec (FLAC)
File wrapper	FLAC (.flac)
Bit depth	24
Sampling rate	96,000 Hz
Number of audio channels	2 (left + right discrete)

Reproduction Details

• No playback EQ curves for cylinder preservation masters. Must be transferred “flat”, with cartridge switched to “vertical”.
• At the start of the preservation master file, a 1kHz tone at operating level (-16 dBFS) for 30 seconds should be added.
• Equipment must be approved by NYPL in writing. NYPL prefers use of Archeophone, Endpoint, or Levin CPS1 cylinder reproducers. Period equipment is not acceptable for use due to age and inconsistency of these machines and the greater potential for damage to the Archival Objects.
• An appropriate assortment of stylus sizes and shapes for cylinders should be on-hand at the vendor transfer station and the best sounding size/shape stylus used for transfer should be noted in the phonoCartridge metadata fields.
• Player setup and playback practices should adhere to best practices.
• Careful determination of the groove pitch (Threads Per Inch or TPI), speed (with correction if needed), and cylinder material (brown wax, Gold Moulded, etc.) need to be determined BEFORE transfer and noted in metadata digitization process notes.

Edit master file specifications: all audio groups

Attribute	Specification
Audio data encoding	Free Lossless Audio Codec (FLAC)
File wrapper	FLAC (.flac)
Bit depth	equal to preservation master
Sampling rate	equal to preservation master
Number of audio channels	Mono, 1; Stereo, 2

Head and Tail Edits (Trimming)

• The “needle-drop” and “needle-lift” present in preservation master files must be edited out of the edit master files.
• Unrecorded portions of the collection object captured in the preservation master shall be eliminated.
• Test tones and any equipment noise at the start and/or end of audible content (such as equipment on/off “clicks” or a stylus in the groove) should be trimmed. Trimming should not result in an abrupt start and/or end of audible content.
• Elimination of the 5-second overlap included on any Preservation Masters that have been split out into multiple files for separate regions / streams / etc.

Level adjustment

• When balance and/or overall level are insufficient a peak level adjustment of max. -2db may be implemented as necessary.

Channel Adjustment

• Ensuring that "mono" is true mono

DATA MEDIA

Deliverables

For each collection object, the following shall be produced:

• One or more preservation master file(s)
• One metadata file per media file
• Image files as described

Capture tools

Preservation masters must be captured/encoded as ISO9660 Disc Images.

Preservation master file specifications: Data optical disc Media in Data Group 1 are distinguished from media in Audio Group 3 and Video Group 3 as follows:

• Data Group 1 uses an ISO 9660 file system to encode data
• Data Group 1 discs do not contain a top-level directory title “AUDIO_TS” or “VIDEO_TS”

Attribute	Specification
File system	ISO 9660 / UDF
File wrapper	ISO (.iso)
Other characteristics	(Same as source)