CalliPhonie Calliphonie Fr En

Résumé Fr En

Content and technical data:From Ref. 1Two speakers (a female and a male, native speakers of French) recorded the corpus. They produced each sentence according to two different instructions: (1) emphasis on a specific word of the sentence (generally the verb) and (2) interrogative intonation. The purpose was to obtain varied intonation patterns. The sentences were recorded in a recording booth, and directly digitalized on a computer (44.1 kHz, 16 bits), using an AKG C414B microphone placed at 40 cm from the speaker’s mouth. All sentences of the corpus were then analyzed in order to extract their fundamental frequency (in semitones), syllabic durations, and intensity.From Ref. 2These experiments are based on a dedicated corpus constructed on 18 sentences, ranging from 1 to 9 syllables length (cf. table 1). Each sentence was recorded in its lexicalized version, and also in a delexicalized version, replacing each syllable by the same /ma/ syllable, in order to obtain reiterant speech [8]. When constructing the corpus, words were chosen with respect to two criterions (use of CV syllable structure and no plosive consonant at the beginning of the words), in order to obtain easily comparable prosodic patterns amongst the sentences and to avoid important micro-prosodic effect due to plosive bursts.Two speakers (a female and male, native speakers of French) recorded the corpus. They have to produce each sentence in a random order, and according to three different consigns: (1) using of a declarative intonation, (2) performing an emphasis on a specific word of the sentences (generally the verb) and (3) using an interrogative intonation. The speakers were instructed to read the sentence and then to produce it using the current intonation style. Once the sentence is recorded in its lexicalized version, they have to reproduce it by using the same prosody, but in its reiterated version. Speakers were able to make as many trials as needed in order to obtain a satisfactory pair of sentences.108 sentences were thus recorded and directly digitalized on a computer (41kHz, 16bits) for each speaker, using an USBPre sound device connected to an omnidirectional AKG C414B microphone placed 40 cm to the speaker mouth, and performing a high-pass filtering of frequency under 40Hz plus a noise reduction of 6dB.Table 1: The 18 sentences of the corpus, from 1 to 9-syllable length (from Ref. 2).Nb syllable Sentence Phonetic1 Non. [nɔ̃]1 L’eau. [lo]2 Salut. [saly]2 J’y vais. [ʒi vɛ]3 Répétons. [ʁepetɔ̃]3 Nous chantons. [nu ʃɑ̃tɔ̃]4 Marie chantait. [maʁi ʃɑ̃tɛ]4 Vous rigolez. [vu ʁigole]5 Marie s’ennuyait. [maʁi sɑ̃nɥijɛ]5 Nous voulons manger. [nu vulɔ̃ mɑ̃ʒe]6 Marie chantait souvent. [maʁi ʃɑ̃tɛ suvɑ̃]6 Nicolas revenait. [nikola ʁəvənɛ]7 Nous voulons manger le soir. [nu vulɔ̃ mɑ̃ʒe lə swaʁ]7 Nicolas revenait souvent. [nikola ʁəvənɛ suvɑ̃]8 Sophie mangeait des fruits confits. [sofi mɑ̃ʒɛ de fʁɥi kɔ̃fi]8 Nicolas lisait le journal. [nikola lizɛ lə ʒuʁnal]9 Sophie mangeait du melon confit. [sofi mɑ̃ʒɛ dy məlɔ̃ kɔ̃fi]9 Nous regardons un joli tableau. [nu ʁəgardɔ̃ ɛ̃ ʒoli tablo]SpeakersThe two speakers are referred to as BD (male) and EF (female); their respective productions are found in two different folders – BD and BDLight, EF and EFLight. The BD and EF folder contains all the recordings, lexicalized and reiterant with Praat TextGrid files that contain their phonemic and syllabic alignments and their glottal closure instants (GCI); the “Light” folders contain only the lexicalized sounds, with their phonetic alignment, GCI and NUCLEI.Labelling and visualizationGlottal closure instants (GCI) GCIs are available for all the sentences and sound files. This was necessary for the PSOLA based pitch modification technique used in the experiments of Ref. 1 and 2.Phonetic labels The Phonetic Labels have been used in a rhythmic evaluation study, reported in Ref. 4.VisualisationThe ProsodicPictures folder contains visualization of the pitch contours and syllabic onsets used in the experiments of Ref. 1, 2, using a stylization procedure described in Ref. 3.

Content and technical data:From Ref. 1Two speakers (a female and a male, native speakers of French) recorded the corpus. They produced each sentence according to two different instructions: (1) emphasis on a specific word of the sentence (generally the verb) and (2) interrogative intonation. The purpose was to obtain varied intonation patterns. The sentences were recorded in a recording booth, and directly digitalized on a computer (44.1 kHz, 16 bits), using an AKG C414B microphone placed at 40 cm from the speaker’s mouth. All sentences of the corpus were then analyzed in order to extract their fundamental frequency (in semitones), syllabic durations, and intensity.From Ref. 2These experiments are based on a dedicated corpus constructed on 18 sentences, ranging from 1 to 9 syllables length (cf. table 1). Each sentence was recorded in its lexicalized version, and also in a delexicalized version, replacing each syllable by the same /ma/ syllable, in order to obtain reiterant speech [8]. When constructing the corpus, words were chosen with respect to two criterions (use of CV syllable structure and no plosive consonant at the beginning of the words), in order to obtain easily comparable prosodic patterns amongst the sentences and to avoid important micro-prosodic effect due to plosive bursts.Two speakers (a female and male, native speakers of French) recorded the corpus. They have to produce each sentence in a random order, and according to three different consigns: (1) using of a declarative intonation, (2) performing an emphasis on a specific word of the sentences (generally the verb) and (3) using an interrogative intonation. The speakers were instructed to read the sentence and then to produce it using the current intonation style. Once the sentence is recorded in its lexicalized version, they have to reproduce it by using the same prosody, but in its reiterated version. Speakers were able to make as many trials as needed in order to obtain a satisfactory pair of sentences.108 sentences were thus recorded and directly digitalized on a computer (41kHz, 16bits) for each speaker, using an USBPre sound device connected to an omnidirectional AKG C414B microphone placed 40 cm to the speaker mouth, and performing a high-pass filtering of frequency under 40Hz plus a noise reduction of 6dB.Table 1: The 18 sentences of the corpus, from 1 to 9-syllable length (from Ref. 2).Nb syllable Sentence Phonetic1 Non. [nɔ̃]1 L’eau. [lo]2 Salut. [saly]2 J’y vais. [ʒi vɛ]3 Répétons. [ʁepetɔ̃]3 Nous chantons. [nu ʃɑ̃tɔ̃]4 Marie chantait. [maʁi ʃɑ̃tɛ]4 Vous rigolez. [vu ʁigole]5 Marie s’ennuyait. [maʁi sɑ̃nɥijɛ]5 Nous voulons manger. [nu vulɔ̃ mɑ̃ʒe]6 Marie chantait souvent. [maʁi ʃɑ̃tɛ suvɑ̃]6 Nicolas revenait. [nikola ʁəvənɛ]7 Nous voulons manger le soir. [nu vulɔ̃ mɑ̃ʒe lə swaʁ]7 Nicolas revenait souvent. [nikola ʁəvənɛ suvɑ̃]8 Sophie mangeait des fruits confits. [sofi mɑ̃ʒɛ de fʁɥi kɔ̃fi]8 Nicolas lisait le journal. [nikola lizɛ lə ʒuʁnal]9 Sophie mangeait du melon confit. [sofi mɑ̃ʒɛ dy məlɔ̃ kɔ̃fi]9 Nous regardons un joli tableau. [nu ʁəgardɔ̃ ɛ̃ ʒoli tablo]SpeakersThe two speakers are referred to as BD (male) and EF (female); their respective productions are found in two different folders – BD and BDLight, EF and EFLight. The BD and EF folder contains all the recordings, lexicalized and reiterant with Praat TextGrid files that contain their phonemic and syllabic alignments and their glottal closure instants (GCI); the “Light” folders contain only the lexicalized sounds, with their phonetic alignment, GCI and NUCLEI.Labelling and visualizationGlottal closure instants (GCI) GCIs are available for all the sentences and sound files. This was necessary for the PSOLA based pitch modification technique used in the experiments of Ref. 1 and 2.Phonetic labels The Phonetic Labels have been used in a rhythmic evaluation study, reported in Ref. 4.VisualisationThe ProsodicPictures folder contains visualization of the pitch contours and syllabic onsets used in the experiments of Ref. 1, 2, using a stylization procedure described in Ref. 3.