Passive acoustic monitoring can be used to assess the presence of vocal species. Automatic estimation of such information is critical for allowing diversity monitoring over long time spans. Among the existing tools, αacoustic indices were originally designed to assess the richness/complexity of terrestrial soundscapes. However, their use in marine environments is impacted by fundamental differences between terrestrial and marine soundscapes. The aim of this study was to determine how they vary depending on the abundance and sound type richness of fish sounds. Fourteen indices used in terrestrial environments were tested. The indices were calculated for files from three sources: a controlled environment (playback of artificial tracks in a pool), in situ playbacks (playback of natural soundscapes), and a natural environment (only natural sounds). The controlled experiment showed that some indices were correlated with the sound abundance but not with the sound type richness, implying that they are not capable of distinguishing the different types of fish sounds. In the in situ playbacks, the indices were not able to capture differences, both in terms of the sound abundance and sound type diversity. In the natural environment, there was no correlation between most of the indices and the abundance of sounds. They were impacted by mass phenomena of biological sounds (e.g., the Pomacentridae sounds in shallow reefs) that cannot inform on fish acoustic diversity. Indices are appropriate when soundscapes are divided into bands. In contrast to terrestrial environments, frequency bands in coastal marine soundscapes do not provide ecologically relevant information on diversity. Overall, indices do not appear to be suitable for inferring marine fish sound diversity.