The Emotional Voices Database: Towards Controlling the Emotion Dimension in Voice Generation Systems

By Adaeze Adigwe, Noé Tits, Kevin El Haddad, S. Ostadabbas, Thierry Dutoit. 2018

111 citations recorded by the source corpus.

In this paper, we present a database of emotional speech intended to be open-sourced and used for synthesis and generation purpose. It contains data for male and female actors in English and a male actor in French. The database covers 5 emotion classes so it could be suitable to build synthesis and voice transformation systems with the potential to control the emotional dimension in a continuous way. We show the data's efficiency by building a simple MLP system converting neutral to angry speech style and evaluate it via a CMOS perception test. Even though the system is a very simple one, the test show the efficiency of the data which is promising for future work.

Subfields: Artificial Intelligence, Experimental and Cognitive Psychology, Generative Models, Speech & Audio, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Experimental and Cognitive Psychology

In Generative Models

In Speech & Audio

In Signal Processing

Citations in the corpus

Neural Codec Language Models are Zero-Shot Text to Speech Synthesizers (2023, positive): “We select prompts from EmoV-DB [83], a dataset containing speech with ﬁve emotions, VALL-E was able to preserve the same emotion from the prompt in speech synthesis, without ﬁne-tuned on any emotional TTS dataset.”
Text-Free Prosody-Aware Generative Spoken Language Modeling (2021, methodology): “Figure A.1 in the appendix illustrates how speaker and prosodic information (emotion) are dis-entangled in raw pitch using a multi-speaker multi-emotion dataset, EmoV (Adigwe et al., 2018).”
LauraGPT: Listen, Attend, Understand, and Regenerate Audio with GPT (2023, methodology): “For the SER task, we collect corpora including MELD (Poria et al., 2018), IEMOCAP (Busso et al., 2008), RAVDESS (Livingstone & Russo, 2018), TESS (Pichora-Fuller & Dupuis, 2020), Crema-D (Cao et al., 2014), Emov-DB (Adigwe et al., 2018), and SAVEE (Jackson & Haq, 2014).”
Emotional Voice Conversion: Theory, Databases and ESD (2021, methodology): “Similar to DES, JL-Corpus (James et al., 2018) consists of emotional speech data from four New Zealand English speakers, and the EmoV-DB database (Adigwe et al., 2018) is built with about 5 h of multilingual speech data covering 5 different emotions and recorded by 4 English speakers and a French…”
EXPRESSO: A Benchmark and Analysis of Discrete Expressive Speech Resynthesis (2023, methodology): “To train the units and vocoder, and to evaluate the results, we use additional open source datasets: LJspeech [11] (LJ), VCTK [12], Librispeech dev-other [13] (LS), Fisher [14] and EmoV-DB (EmoV) [9].”
Textless Speech Emotion Conversion using Decomposed and Discrete Representations (2021, methodology): “, 2013) and EmoV (Adigwe et al., 2018), and stop after 3M”