1 of 57

CRIM Summer School 2021

Calendar

Sessions are 9:00-10:30 AM EST

1 Wednesday July 14, 2021 Introduction and Key Concepts
2 Thursday July 15, 2021 Exploring the Music
3 Wednesday July 21, 2021 Navigating the Notebooks for Data Analysis
4 Thursday July 22, 2021 Data Analysis, Continued

Zoom

Join Via Zoom

Slack

The conversation will continue between sessions via Slack. Sign up for Slack, then join CRIM @ Slack here.

2 of 57

Introduction + Aims

3 of 57

Thanks to . .

Advisory and Editorial

Philippe Vendrix (CESR, Tours)

David Fiala (CESR, Tours)

Philippe Canguilhem (CESR, Tours)

Vincent Besson (CESR, Tours)

Julie Cumming (McGill University)

Peter Schubert (McGill University)

David Crook (University of Wisconsin)

Jesse Ann Owens (University of California)

Jesse Rodin (Stanford University)�

Technical

Daniel Russo-Batterham (Melbourne University, Australia)

Alex Morgan (McGill University, Canada)

Micah Walter (Smith College, USA)

Andrew Janco (Haverford College)

Raffaele Viglianti (MITH, University of Maryland, USA)

Emilio Sanfilippo (CNR, Trento, Italy)

Linh Le (Bryn Mawr College, USA)

Trang Dang (Bryn Mawr College, USA)

Laurie Tupper (Mount Holyoke College, USA)

And many other scholars and students!

4 of 57

CRIM Research Questions

Musical

How did compositional practice change over the course of the sixteenth century?
What connections can we draw between 16th-century ideas about compositional practice and patterns that we observe today?
What relationship does the process of musical modelling bear to the wider humanist concern for imitatio of classical models?
How do the processes of adaptation heard in our corpus compare with those of other musical styles?

Digital

How can digital techniques help us to model new notions of similarity among musical works?
How might the digital domain inaugurate new modes of scholarly communication?

CRIM is about SIMILARITY in music, which we examine through the particular instance of the Imitation Mass, in which a chanson, madrigal, or motet reheard, rebuilt, and transformed as the foundation of a cyclic setting of the five movements of the Mass Ordinary.
In the course of our work, we’re trying to answer a series of related questions--both musical and digital, as summarize here

Changing compositional practice
The relationship between modern and historical conceptions of contrapuntal practice
The place of such pieces in the wider cultural phenomen of imitation
The relationship between Renaissance practices and other forms of musical borrowing--from Handel to hip-hop.
How digital tools can aid any of this, what they tell us, and what we can ask them to do
How to imagine new, collaborative modes of scholarly communication among musicologists, music theorists, and data scientists.

5 of 57

CRIM to Date

33 Masses and their Models

Echoes of Josquin (Févin, Morales, Palestrina)
The Mass ca. 1550 (Sermisy, Gombert, Phinot, Guerrero)
Self Reflections (Lasso, Palestrina, Sermisy)
Proposed Additions (2021-2022)

Analytic Vocabularies
Database of 2500 Relationships, with detailed annotations
Data analysis tools
Conferences, Study Days, Classroom Visits, Assistantships
Web application: http://crimproject.org/
Editor’s Forum: https://sites.google.com/haverford.edu/crim-project/home
All: Open Source, Collaborative

CRIM is about SIMILARITY in music, which we examine through the particular instance of the Imitation Mass, in which a chanson, madrigal, or motet reheard, rebuilt, and transformed as the foundation of a cyclic setting of the five movements of the Mass Ordinary.
In the course of our work, we’re trying to answer a series of related questions--both musical and digital, as summarize here

Changing compositional practice
The relationship between modern and historical conceptions of contrapuntal practice
The place of such pieces in the wider cultural phenomen of imitation
The relationship between Renaissance practices and other forms of musical borrowing--from Handel to hip-hop.
How digital tools can aid any of this, what they tell us, and what we can ask them to do
How to imagine new, collaborative modes of scholarly communication among musicologists, music theorists, and data scientists.

6 of 57

CRIM Masses http://crimproject.org/masses/

CRIM Models http://crimproject.org/models/

7 of 57

8 of 57

Next Steps:

CRIM Summer School: 4 90-minute sessions, July 14-22
2021-2022:

Classroom Visits
Fall School at Padua
Spring-Summer Conferences in Tours, Haverford

2021-2023: Paid Assistantships for Editors and Analysts

9 of 57

Human Annotation and Analysis

10 of 57

Human Annotation http://crimproject.org/relationships/31/

Observation

Model

Musical Type

Score Address

Observer

Observation

Derivative

Musical Type

Score Address

Observer

Relationship

Relationship Type

Observer

11 of 57

One Observation from Josquin’s Benedictus es

Visit http://crimproject.org/observations/61/

12 of 57

Palestrina and The Worlds Longest Fuga: 26 entries! http://crimproject.org/relationships/1407/

Tallis Scholars

Josquin Fuga

Palestrina Fuga

13 of 57

Controlled Vocabularies

14 of 57

Pietro Cerone, El melopeo y maestro (1613)

Contrapuntal Commonplaces

Presentation Types Made Simple

Abbreviations and Rules

15 of 57

Thesaurus of Musical Types

with . . .

PEN exemplified

PEN defined

Abbreviations and Rules

16 of 57

Thesaurus of Relationship Types

Mechanical Transformation

Non-Mechanical Transformation

17 of 57

Key Vocabulary Updates

Musical Types

Cadences Simplified

Irregular (which voice?)
Dovetail (which voice? Above/Below/Between the cadential pair?)

Homorhythm

now includes Dialogue (for patterns in which there are no repeating modules

Interval Patterns

now includes Dissonance (prepared/unprepared/suspended)

Form and Process no longer used.

18 of 57

Key Vocabulary Updates

Relationships Types

Mechanical and Non-Mechanical Transformation with Common Sub-Types

avoids confusion for analysis

Self Relationships (allows any work to be its own model-derivative)

Replaces “Form and Process” from Musical Types
Enchainment (when a counter soggetto or fragment of a previous melody becomes the heatmotif of a new soggetto)
Repetition (direct repetition of some passage, perhaps varied)
Return (related material comes back later in that movement or piece)

19 of 57

Remember Best Practices

Musical Types are Mutually Exclusive

complex forms always supercede simple ones

If Contrapuntal Duo, do not mark Soggetto and Counter-Soggetto
If Non-Imitative Duo, do not mark Contrapuntal Duo
If Fuga, do not mark Soggetto

musical types should not be combined

with the except Cantus Firmus and Interval Patterns
on the other hand: a passage cannot be ID and a fuga at the same time Don’t do This or This or This

the same passage can often be understood in different ways

some fugas contain hidden PENs (or IDs)
consider marking core structure, and then mention added entires in a comment (but not in the selection or metadata)

20 of 57

On the Analyst’s Desk:�Markup Scores Keep Index of Relationships

Introduction

In the CRIM Project we aim to make detailed observations about relationships between masses and their models. We’ll build our database of observations in steps:

analysis of mass and model, using Thesaurus of Musical Types
map relationships between mass and model, using Controlled Vocabulary of Relationships
creation of EMA Citations (directly from scores)--this will be later!
data entry, in which EMA Citations and Relationships will be added to the database via webform. This will be later!
discussion of results, via web application (this will is still in development, but the CRIM Omeka site will provide a basic hub for our work)

21 of 57

On the Analyst’s Desk:�Markup Scores Keep Index of Relationships

Survey the Model and Mass

With score, make note of the important structural features of the piece, including

which Presentation Types?
which Cadences? What tones and types?
notable text-tone relationships? These can be in comments you keep separately.
Mark these observations on the score, using our standard abbreviations and symbols, and transfer these to a worksheet for reference. You will need these during Data Entry.

You can use the text of the piece in CRIM as a guide!

Remember: you will need to be selective, focusing on the patterns that are most interesting from the standpoint of borrowing and transformation.

22 of 57

On the Analyst’s Desk:�Markup Scores Keep Index of Relationships

Mark RELATIONSHIPS between Mass and Model

As you find relationships, note them in the Mass Score. We suggest putting these in a square box (vs the circles for the previous mark-up). Or write these relationships in RED.

If you use #’s for Musical Types, then use Letters for Relationships. “Relationship A uses Model #1 and Mass Kyrie #4”

Keep a separate (typed) table of Relationships to which you can refer when it comes time for data entry.
Don’t worry if the same passage in the model seems to be used in more than one Relationship, and in different ways!
Don’t worry if one passage in the Mass seems to have multiple kinds of connections with the model! We can (and indeed, should) consider these as separate relationships.

The database and heatmaps will help us discover patterns that take place in the same section of a piece, or involve the same kind of transformation!

EMA citations will as exact as you wish (via browser selection). These will be done later!

23 of 57

On the Analyst’s Desk:�Markup Scores Keep Index of Relationships

View Sample Marked Scores and Relationships

Note that these use the OLD Musical Types and Relationship Types, but show some ways keep track of insights on the score and in a worksheet.
Model 8 (Josquin, Ave maria)
Mass 5 (Fevin, Missa Ave maria) Gloria
Mass 5 (Fevin, Missa Ave maria) Agnus dei
Relationships between Model 8 and Mass 5
Freedman Marked Models (Rore, Lupi, Josquin, Palestrina, Sandrin, Sermisy)

24 of 57

Exploring the Music

25 of 57

The same soggetto should be heard at the start of each movement of the Mass, but in different contrapuntal treatments, for instance, by having the soggetto appear in different voices of the Mass, or by having the surrounding parts enter at different time intervals around it.
The endings of the various subsections of movements (like the last Kyrie, the subsections of the Gloria, Credo, etc) should also resemble each other, and make use of the main soggetto. They should also be varied in terms of contrapuntal treatment or accompaniment of the soggetto.
The entire Agnus dei movement is often for a larger number of voices than the remainder of the Mass. The third Agnus should also use the main soggetto of the model.
The Christe should make use of a subsidiary soggetto of the model.
The Gloria and Credo will be shorter and clearer in terms of the treatment of the text than the Kyrie, Sanctus, and Agnus dei, which can be elaborate.

Advice from Pietro Pontio

Ragionamento di musica (1588)

26 of 57

Motet Text Phrase	Motet measures	Kyrie	CRIM Relationships
Veni sponsa Christi	1-22	Kyrie I: 1-9	R 1037
accipe coronam	19-36	Christe: 20-44	R 291
quam tibi Dominus	36-52	Kyrie II: 45-55	R 1155
praeparavit in aeternam.	50-67	Kyrie II: 55-67	R 309

Palestrina’s Veni sponsa Christi and his Mass

27 of 57

Veni sponsa Christi: The First Soggetto of the Motet and its Treatment in the Mass

As Observed by CRIM participants

Motet mm. 1-7 Imitative Duos C>A>T>B

Selected Relationships (see complete list for this phrase)

Mass Movement and Musical Type	CRIM Relationships?
Kyrie m. 1 ID CA >TB	R 260 R 104 R 1037
Credo m. 1 ID CA > TB, with new gap between entries.	R 2496
Sanctus m. 62 ID TB > CA Hosanna in triple mensuration	R 1531
Sanctus m. 82 PEN a3 T>B>C	R 1541

28 of 57

Veni sponsa Christi: The First Soggetto of the Motet and its Treatment in the Mass

As Observed by CRIM participants

Motet mm. 1-7 Imitative Duos C>A>T>B

Selected Relationships (see complete list for this phrase)

Mass Movement and Musical Type	CRIM Relationships?
Gloria m. 44 PEN a4 T>C>B>A	R 265 R 333
Sanctus m. 1 NIM CT > AB > TB	R 346
Agnus dei m. 35 Fuga a3 C>A>Q	R 403
Sanctus m. 59 Fuga a2 A>C Hosanna in triple mensuration	R 1529
Credo m. 53 Homorhythm, with soggetto as tenor for Et incarnatus est	R 1730

29 of 57

Veni sponsa Christi: The First Soggetto of the Motet and its Treatment in the Mass

As Observed by CRIM participants

Motet mm. 1-7 Imitative Duos C>A>T>B

Selected Relationships (see complete list for this phrase)

Mass Movement and Musical Type	CRIM Relationships?
Gloria m. 44 PEN a4 T>C>B>A	R 265 R 333
Sanctus m. 1 NIM CT > AB > TB	R 346
Agnus dei m. 35 Fuga a3 C>A>Q	R 403
Sanctus m. 59 Fuga a2 A>C Hosanna in triple mensuration	R 1529
Credo m. 53 Homorhythm, with soggetto as tenor for Et incarnatus est	R 1730

30 of 57

Examples

^{What did composers hear? Did Pontio reflect their practice?}

^{What did CRIM Analysts observe? What can you observe?}

Févin's Missa Ave Maria, based on Josquin's motet.

View Marked Scores View Selected CRIM Relationships

Palestrina’s Missa Veni sponsa Christe, based on his own motet.

View Marked Scores

Guerrero’s Missa Sancta et immaculata virginitas, based on Morales’ motet

View Marked Scores

Guerrero’s Missa Super flumina Babylonis, based on Gombert’s motet

View Marked Scores

Manchicourt’s Quo abiit dilectus tuus, based on his own motet

View Marked Scores

31 of 57

Examples

^{What did composers hear? Did Pontio reflect their practice?}

^{What did CRIM Analysts observe? What can you observe?}

Lasso’s Missa Ite rime dolenti (based on Rore’s madrigal)

View Marked Scores

Lasso’s Missa Susanne un jour, and Pere Riquet’s Missa Susanne un jour, based on Lasso’s (and Lupi’s) chanson

View Marked Scores

Sermisy’s Missa Tota pulchra es, based on his own motet

View Marked Scores

Phinot’s Missa Si bona suscepimus, based on Sermisy’s motet

View Marked Scores

32 of 57

Music Encoding and Structured Data

33 of 57

33

As Title

<title>Hamlet</title>

As Speaking Role

<w xml:id="fs-ham-0620180">HAMLET</w>

</speaker>

As Geographical Entity

<settlement type="hamlet">Blaise Hamlet</settlement>, <region type="district">Henbury</region>

</placeName>

TEI: Logical Disambiguation of “Hamlet”

34 of 57

34

TEI: Graphical and Logical, from The English Broadside Ballads Archive (UCSB)

35 of 57

TEI <choice>

Logical and Graphical

CESR BVH Project

36 of 57

Query finds _all_ abbreviations and expansions in one TEI file (or many!)

37 of 57

Sibelius

MEI

Verovio

Editor’s Desk

User’s Screen

View MEI File

Symbolic Scores ⇔ Logical Encodings

Humans read “symbolic” scores, in which pitch, duration and other representations of sound take graphical form.
Machines read “logical” ones, in which the various dimensions of musical space and time are presented as numbers and characters.
There are many such digital standards for music information (like MIDI, or MusicXML).
The Music Encoding Initiative (MEI) is the one we use in CRIM on account of it’s rich capacity to represent the full apparatus of scholarly markup (musica ficta, variants) in digital form. It’s also an open-source, community based standard (unlike MusicXML, which is fact proprietary)
We created MEI scores starting with Sibelius or Finale, and then use the MEI to render scores dynamically in the user’s browser.
This means we can instantly create examples, slices, highlighted passages for comparison. And analytic annotations created by users can be instantly transformed from clickable notes on their screen into data for further analysis.

38 of 57

MEI: One Bar, with Editorial Accidental

38

View MEI File

39 of 57

MEI + EMA (Enhancing Music Addressability)

39

EMA Reference= filename + 8-9/1,1/@all,@all

measures/staves/beats

Read the full MEI specification here

40 of 57

Structuring and Storing Data: Defining Object Types

40

Observation 333

creator: CRIM_Person_0103�piece: CRIM_Model_0014�ema: 1/2/@3-4�soggetto: true�soggetto ostinato: true�created: 2019-11-11

Person CRIM_Person_1003

name: David Fiala�dates: [empty]

Piece CRIM_Model_0014

composer: CRIM_Person_0021�genre: motet�title: “Si bona suscepimus”

pdf_link: https://crimproject…�mei_link: https://crimproject…

41 of 57

Accessing the Data: Django Views

A view translates the data into a consumable format, following�any links that are needed to show the user what is desired.

Two renderers:

Human-readable: https://crimproject.org/observations/4103/
Machine-readable: https://crimproject.org/data/observations/4013/

41

42 of 57

Practice Making EMA Selections

The CRIM Database is not accepting new Observations/Relationships while we complete some updates.

But you can see how the EMA selection tool works here.

Select a piece from the link at the left
Click on any combination of notes
Select a region
Be sure to include stems
Avoid edge of screen
Command (or right click) and Click to de-select a note from the passage

43 of 57

Data Analysis: CRIM Intervals

44 of 57

CRIM Intervals Project: Machine Score Reading and Analysis

MEI (encoded scores)
music21 (a Python library for music analysis), by Michael Cuthbert
CRIM Intervals: (Pandas Python library to find melodic, harmonic, and contrapuntal patterns)

What Can it Do?

CRIM Tones, nGrams, and Modules (What distribution of notes and patterns in a piece?)
CRIM Classifiers (Where are the Cadences, Fugas, IDs, NIms, and PEns in a piece?)
CRIM Data Viewers (What have analysts observed about CRIM works?)
CRIM Heat Maps (Where in a given piece are the Relationships or Observations?)
CRIM Network Tools (What “communities” of pieces can we identify?)
CRIM Similarity in Time (What can statistical method tell us about patterns in time?)

Adjustable

Diatonic/Chromatic, Compound/Simple, etc
Real Durations, or by Increments
Degree of Similarity
Many other features . . .

45 of 57

CRIM Tones, nGrams, and Modules (What distribution of notes and patterns in a piece?)

45

46 of 57

CRIM Tones, nGrams, and Modules (What distribution of notes and patterns in a piece?)

combinations of melodic soggetti, and in particular how they intersect to make repeating ‘modules’ of vertical intervals. We can now do this, too.

46

12 10 8 8

5 3 1 1

47 of 57

CRIM Tones, nGrams, and Modules (What distribution of notes and patterns in a piece?)

Overlapping N-Gram of 5 elements, representing three ‘states’

Vertical Interval 1 > Tenor Motion 1 > Vertical Interval 2 > Tenor Motion 2 > Vertical Interval 3

48 of 57

CRIM Classifiers (Where are the Cadences, Fugas, IDs, NIms, and PEns in a piece?)

Compare with human observations: http://crimproject.org/relationships/2597/

49 of 57

CRIM Metadata Viewer

Search by Piece, Musical Type, Relationship Type
Download Data as CSV, Charts
https://crimdataviewer.herokuapp.com/

.

49

50 of 57

CRIM Heatmaps

.

50

51 of 57

CRIM Network Graphs

.

51

52 of 57

Try CRIM Intervals Yourself, via this link

Click “Launch Binder”

53 of 57

Try CRIM Intervals Yourself, via this link

Wait for Binder to launch, then
Open “binder”

54 of 57

Try CRIM Intervals Yourself, via this link

Open “1_CRIM_NB”: Pattern Search

or

Open “2_CRIM_NB”: Heatmaps and Networks

55 of 57

Try CRIM Intervals Yourself, via this link

Open “Table of Contents”

Navigate to sections
Shift+Enter to run a cell

56 of 57

CRIM Intervals allows the user to adjust the following:

Chromatic or Diatonic intervals
“0” or “1”- based indexing
Length of the Vectors (minimum number of notes to match)--rests are boundaries
Exact or Close Matches (including how many ‘changes’ are allowed between matches)
Real or incremental durations (by every half-note, quarter-note, etc)
How many matches are required before they are reported
Temporal distances among entries in contrapuntal presentation types
Filter according overall length of soggetti
How many and which pieces you search at once

57 of 57

Results include (in one or more pieces)

A list of vectors and durations for each match
Sum of durations for each soggetto
The ratios of the durations (so that we can find diminution, augmentation)
The piece, voice part, measures and beats
The EMA address of the passage (which will then render in Verovio)
Predicted Presentation Types of Contrapuntal Soggetti
Melodic intervals of all soggetti, expressed as N-Grams
Contrapuntal intervals between any pair of voices, expressed as N-Grams