TECHNOMALIST
Adobe has promoted three AI audio tools out of beta on its Firefly platform, giving creators and marketers browser-based access to music, sound effects, and speech generation. TechRadar Pro's reviewer tested each tool and found that while none will replace professional audio work soon, the music generator in particular delivered usable results for everyday video production.
Generate Music lets users either type a free-form prompt or upload a video so the AI can infer an appropriate style. In one test, the reviewer fed a 27-second clip from a product review. Firefly suggested a prompt for a calm, ambient electronic track suited to a product demonstration. The system generated four variations in a couple of seconds. The initial results were described as serviceable but somewhat generic; adding more descriptive terms—such as jazz, cinematic, vlog, and trailer—and changing the energy level from low to medium produced a noticeable improvement. The reviewer noted that, as with other generative tools, richer prompting yields better audio. Users can download the music alone or paired with the video.
Generate Speech offers a choice of models: Adobe's commercially safe Firefly option or the ElevenLabs Multilingual V2 partner model. The interface presents 45 speaker profiles spanning different genders, ages, and voice types. After pasting text, the system can auto-detect language, though manual selection is available. Users can insert pauses and additional lines. The reviewer found a workflow issue: previewing requires selecting the entire text and using a context menu rather than a simple play button, which felt like an unnecessary obstacle. Pauses had to be adjusted to make the reading sound natural, and the word 'land' was mispronounced as 'lan' during testing.
Generate Sound Effects requires manual description of the desired sound, with an option to let the AI expand the prompt. When the reviewer tested a prompt for a sliding whistle, the AI produced four ear-piercing variations that were unusable. Switching to a request for crickets in a field yielded a more pleasant result. The tool supports multiple audio tracks on a timeline, with handles for positioning, volume, and deletion—making it feel closer to a simple audio editor.
Overall, the reviewer concluded that the tools need UI refinements for faster workflows and will not displace professional sound designers, composers, or musicians. Speech has moved far beyond robotic early text-to-speech, but nuanced delivery still favors a human performer. Sound effects were judged useful and, in the reviewer's listening tests, indistinguishable from traditional effect libraries. Generate Music emerged as the strongest feature: even base prompts produced audio suitable for product demos, reviews, or business explainers, and output improved with each generation.
The launch follows an Adobe survey conducted with Berklee College of Music, which found that 32.7% of video creators, musicians, and marketers use AI-generated music as the final track in their published work. The reviewer sees that statistic as a likely driver for this update, aimed at speeding up creative workflows for marketers and everyday video production.
See an error? Read our corrections policy or email [email protected].
