ChatGPT Reads the Lyrics, Not the Tune
When you talk to ChatGPT it feels like it is listening. I wanted to know what actually arrives on the other side, so I picked the simplest thing a voice carries besides words: a melody. I hummed Happy Birthday to ChatGPT, in every way the app lets me speak to it, and then ran the same question at scale through the voice models OpenAI sells to developers. Two buttons, two different things The…
The brief examines the difference between two voice inputs in OpenAI's ChatGPT: Dictation and Voice. Dictation, which uses the small microphone, transcribes spoken words into text before processing them, while Voice, which uses the Voice button, receives the audio directly. When the author hummed "Happy Birthday" to ChatGPT, Dictation remained silent, as there were no words to transcribe.
However, Voice acknowledged its inability to identify songs from humming or melodies, mentioning that they are working on this feature. To further investigate the capabilities of OpenAI's voice models, the author tested three models via the API using recordings from the MTG-QBH dataset. While the text models performed similarly to Dictation in recognizing songs with lyrics, they were unable to identify songs without lyrics.
The Voice model remains limited in its ability to process melodies, as it does not yet have the necessary functionality to identify songs from humming or melodies.
Brief written by urgent.news from Dev.to's own syndicated text. Machine-written; may contain errors — check the original.