Voice and Singing in
Popular Music in the U.S.A. (1900–1960)
vocalmetrics
A web-tool for exploring the popvoice
Concept and implementation
Martin Pfleiderer, Tobias Marx, Tilo Hähnel, Inga Langhans (HfM Weimar)
Rainer Groh, Axel Berndt, Felix Schönfeld, Eva Brumme, Huong Nguyen (TU Dresden)
Version 1.0 (01.07.2013)
Criteria of selection
The audio samples are a selection of several genres of popular music in the United States between 1900 to 1960: Blues, gospel music, vaudeville, jazz, American popular song, hillbilly/country music, rhythm'n'blues and rock'n'roll.
The database is currently supplemented.
The audio samples are supposed to represent typical vocal features of important artists in popular singing.
Criteria of selection have been:
- Phrases and passages should represent typical personal styles, timbres, or other vocal features within a time-frame of ten to twenty seconds.
- Covering of an artist's multi-facetted singing and vocal development over a longer time period.
- High sound quality or audibility of singing (recordings with as little noise and as little accompaniment as possible).
- title: name of the song which contains the selected passage
- artist: name of the vocalist
- year: year of recording or publication
- genre: genre affiliation of the song
- label: recording label or record company which published the recording first; relevant is the sub-label, not the major-label beeing owner at the time
Ratings of vocal features
The ratings of the nine vocal features were inspired by the Cantometrics project of Alan Lomax [1,2,3].
All the audio examples have been rated by Inga Langhans, Tilo Hähnel and Tobias Marx independently.
Rating scales with five levels were used.
However, some ratings are compositions of two partial ratings (see below).
Details for the nine features are listed below. Separate ratings of frequency or intensity of a vocal feature were used in glissando, vibrato, roughness, breathiness, register crossing and off-beat frequency. The meaning of values between 0 and 4 are:
|
4 3 2 1 0 |
very strong or very frequent quite strong or quite frequent neutral quite weak or quite rare very weak or not existing or very rare or not existing |
| Feature | Description | Audio sample | ||
|---|---|---|---|---|
| Glissando | Decisive for the rating has been the subjective perception of obvious gliding between pitches or the perception of pulling and dragging of tones. Glidings over long–time periods as well as glidings over high frequency intervals contribute to the experienced intesity. The overall rating contains partial ratings of glissando intensity and glissando frequency. |
strong gliding
rare gliding
|
||
| Vibrato | Decisive for the rating has been the subjective perception of periodical changes of pitch and intensity within a frequency frame of 4–9 Hz with no regard to the measurable dimensions of frequency and amplitude. The overall rating contains partial ratings of vibrato intensity and vibrato frequency. |
strong vibrato
rare vibrato
|
||
| Roughness | Decisive for the rating has been the subjective perception of jitter, shimmer and roughness as can be veryfied by additionally emerging subharmonics in the spectrogramm [4,5]. The overall rating contains partial ratings of intensity and frequency of percieved roughness. Roughness is evaluated in dissociation of breathiness (see next feature). |
strong roughness
no roughness
|
||
| Breathiness | Decisive for the rating has been the subjective perception of noise caused by air swirls in the vocal tract. The overall rating contains partial ratings of intensity and frequency of percieved breathiness. |
strong breathiness
weak breathiness
|
||
| Articulation | Ratings of enunciation from »incomprehensible» (0) to »very clear« (4). In passages without lyrics (e.g. scat singing) decisive for the rating has been the clarity of articulation. |
clear enunciation
weak enunciation
|
||
| Intensity | Decisive for the rating has been the subjective perception of volume – not the measurable level of loudness. The overall rating contains partial ratings of intensity of singing between very »soft« (0) to »very loud« (4) as well as variability of intensity between »none« (0) to »very high« (4). |
high intensity
low intensity
|
||
| Vocal register crossing | Decisive for the rating has been the subjective perception of counterbalanced or emphazised vocal registers. Emphazised crossing of vocal register (e.g. in jodeling) are rated as strong. Not existing or not perceptible crossing of vocal register are rated low or as none existing. The overall rating contains partial ratings of intensity and frequency of percieved crossing of vocal register. |
emphazised change
of vocal register
no change of
vocal register |
||
| Tempo Rubato |
As an exception, a bipolar rating scale was used for strict timing or intense rubato in relation to the meter of the accompanying music.
Values are: 4: very strong tempo rubato; 3: quit strong tempo rubato; 2: neutral; 1: quite accurate timing; 0: very accurate timing |
very strong
tempo rubato
very accurate
timing |
||
| Off-beat Frequency | Frequency of emergence of off-beat accents. Off-beat timing means singing tones early on positions before a beat. Therefore off-beat timing has no relation to accurate timing or tempo rubato. |
strong off-beat timing
no off-beat timing
|
Calculation
-
The partial ratings of strongness and frequency in glissando, vibrato, roughness, breathiness, register crossing and off-beat frequency are combined as follows:
When rtotal is the overall rating from both partial ratings rpartial, and a is the parameter for weighting, then:
Literature:
Music:
last update: 20.10.2014