Test methods for Score-Based Interactive Music Systems Toward a formal Specification
2
0
0
Texte intégral
(2) Test methods for Score-Based Interactive Music Systems Clément Poncelet. Florent Jacquemard. ([email protected]). ([email protected]). How be sure of a Score-Based Interactive Music System’s behavior during shows ? Ensuring a “guarantee” on the behavior of a complex real-time system is not trivial and requires formal method. Today, Antescofo’s behavior is “manually” checked by the compositor during rehearsals. Since rehearsal performances represent just a few set of all possible interpretations of a score, this method is not rigorous.. INRIA Mutant Project Team - (UMR-9912) IRCAM - 1 Place Igor-Stravinsky 75004 Paris. Time Line. Manual piece preparation Mixed score. Composition time. Performance/Show time. ≡. Listening. verification. requirements. musician. Input Audio. Score follower. s ? u o d r e o k g i c r e h C very t o N. audio softwares. Output Messages. Compiling Score into Intermediate Representation. MAX / PD. Covering the set of possible interpretations. Toward a formal Specification sn. Models of groups s0. ḡ?. α1!. s1. tei+1. αn !. s2. Model of group trigger. with “loose” synchronization strategy s0. sm. ǧ?. s1. d. α1!. s2. d. s3. sm. αn !. sn. tei+1?. tei ?. sn. s0. s3. ?. s1. αn !. s2. so. sn. αn!. s0. ǧ?. d. s1. s2. α1 !. s3. d. sm. αn!. sn. te. g?. tei+1?. so. group1?, t:=0. e2 ?,. 0. S05. a11!. e t a i d e m n r o e i t In resentat Rep. S07. C e¯1 ?, t:=0. S08. ¯ group1?, t:=0. C. S03. a11!. S09. a12!. t=10. s2. e2 ?. s3. te2 !. ai Output messages. te1 ! sm. sk. te1..n−2 !. sn. ten−1 !. The Intermediate Representation is a step toward a formal specification of Antescofo’s expected behavior. It is a model of How Antescofo’s reactive engine should compute its timed output for every inputs (i.e musician performances). IR can be represented with Extended Timed Input/Output Labelled Systems. Each system starting at its s0 or s¯0 state and waits on an edge δ? for a fire of δ! of the symbol δ. IR is able to specify multi-clocks, concurrences, dynamic variables... Outputs ai. e1 · 0. note1 ?. S01. e1 !. S02. note2 ?. S03. model. e2 !. s00. S04. e1 !, t:=0. S01. e2 !, t:=0 t≥min. S02. t≤M AX. C ¯ group1!. S01. note2 ?. S04. e1 !. System. Environment. S10 group2!. S01. t=10. e2 ?, t:=0. S00. t := 0. S01. a2! t=50. S02. d e Tim n o t a m o t u A. e2 .5 · e. e1 · 0. 2. .33 · e. Some g 2 enerat ed input t races. Existing tools for timesting real time systems, with an automatic test cases generation, are based on the model of network timed automaton (TA). IR arez translated into TA straightforwardly, at the prive of expressiveness loss. A set of artificial musician performances is automatically generated, with a cover guaranty (i.e quality of tests). The conformance of real traces to the formal model allows us to increase the guarantee on the stability of Antescofo at execution time.. S02. S04 group2?. sn. s4. en ?. Killable. s3. a2 ?. e2 !, t:=0. C. C. s1. te1 !. e3 !. s2. Observable I/O. S02. S00. S03 a11!. group1!. e2 !. 1.0. S00. =0. =0. C. e1 ?, t:=0. t:= 0 a12!. e1 ?. s1. k!. s1. Environment model. a1 ?. Inputs ei S01. S06. , t: e¯2?. , t: e¯2?. S00. S02. e1 !. tei+1! tei+1! αn−1! sj s2 sn. : commited states e2 ?, t:=. s0 [d1..d2]. e2 ?. so. Model Based Test Generation S01. sn. αn !. so. 1?. (i.e Timed Automaton used by the Uppaal verification suite for realtime systems). a11! t=40. αn!. sm. s0. Translating into standard formal models. C. α1 !. s1. i+. tei+1?. d. tei+1?. i +1. um. s3. Event proxy model. ?. αn!. α1!. α1!. sn. te. u2. ḡ?. tei+1? 1?. um. 1?. t e i+. u2. α1!. ? t e i+1. 1?. sm. 1 +. i+. ǧ? s0. s2. αn!. t e i+. te. d. tei. s1. α1!. α1!. s2. sj. tei ?. Models of groups with “tight” synchronization strategy d. d. s1. Input events ei. e2 !. S00 e¯2 ?, t:=0. S03. Artificial rehearsals. Timed Automaton Network. s c e t of n. A show !. s ec u MUTANT website. the. o. Generated input trace. coVer. Real output trace. KO. red for. Test cases. Traces suite generation and translation. Tempo Curves. Check conformance Time functions. OK. DGA. Expected output trace. Équipe - Projet Mutant. UMR-9912. ANR Inédit. Project ANR-12-CORD-0009. UPMC. CNRS. Fuzzing.
(3)
Documents relatifs
We investigate timed automata based models, with formally defined semantics, in order to study the behavior of a given system with some imposed timing constraints when deployed in
université du Havre, Kevin Sutton, université de Grenoble, Julien Thorez, CNRS, Pierre Thorez, université du Havre, Jean Varlet, université Savoie Mont Blanc,
We present a formal model (Variable Driven Timed Automata) adapted for such systems in- spired from timed automata using variables as inputs and outputs, and clocks.. In this model,
Finally, the fourth approach is an automated, model-based approach that reuses field data to generate test inputs that fit new data requirements for the purpose of testing
Influence of silver nanoparticles on fibronectin adsorption at the surface of a plasma-deposited organosilicon coating Laurine Martocq * , Pascale Chevallier 1 , Morgane Laurent 1
Figure 4 REMOTE OCULOMETER ELECTRO-OPTICAL UNIT.. a) The Oculometer eye-tracking/control programs were loaded into the mass core memory from the disc memory unit. That
A regularization algorithm allowing random noise in derivatives and inexact function val- ues is proposed for computing approximate local critical points of any order for
We have taken a traditional, mature course in paradigms of computer programming and we have recast it in a new course containing two tracks: a SPOC (Small Private Online Courses)
