ACE Studio’s new ACE Music v1 model topped Suno v6 in a blind listening test, scoring 1,160 Elo points and showing stronger melodies and cleaner song endings.
ACE Studio’s flagship AI music model, ACE Music v1, has taken first place on Artificial Analysis’ vocal music leaderboard, surpassing Suno’s widely‑used v6 model. In a double‑blind listening test that measured listener preference with an Elo‑style rating, ACE Music v1 earned 1,160 points compared with Suno v6’s 1,144, marking a narrow but notable lead. The benchmark, AA‑Music‑Vocal v1.1, evaluates the quality of AI‑generated songs that include vocals, with evaluators hearing paired tracks generated from identical prompts and selecting the one they prefer.
According to Decrypt’s hands‑on assessment, ACE Music v1 consistently produced more melodic and coherent compositions in mainstream styles. The vocal output was described as “more natural,” with fewer synthetic artifacts such as mispronunciations, excessive autotune, or hallucinated words. The model also handled song endings more cleanly, avoiding the strange sounds and abrupt changes sometimes heard in Suno‑generated tracks. However, Suno retained an advantage in niche genres; in a melodic death‑metal test, Suno delivered more stylistically appropriate aggressive vocals, while ACE’s output featured a clean female voice that did not match the requested intensity.
ACE Studio emphasizes a songwriting‑assistant workflow. Its “Idea” feature expands a basic concept into full lyrics, song sections, and arrangement suggestions, allowing users to refine rather than restart each generation. The v1.1 preview adds a scoring step, generating a musical score before audio, and lets users edit melodies, chords, and export MIDI or isolated vocal tracks. While the flagship ACE Music v1 model remains proprietary with access limited to the company’s web portal, the firm has released ACE‑Step 1.5, an open‑source family of smaller models that can run locally. ACE also announced plans to open‑source a “Flash” version of the flagship model later in 2026, aiming to balance quality, speed, and hardware requirements.
The upcoming release of ACE Music v1 Flash will be a key test of whether the performance gap with Suno can be narrowed for users who prefer on‑device generation. Observers will also monitor whether Suno updates its vocal model to address the weaknesses highlighted by the benchmark. Finally, the broader AI‑music market may see increased competition as more open‑source tools emerge, potentially shifting the balance between proprietary services and community‑driven projects.
The Elo rating reflects listener preference in a blind test; a higher score means evaluators chose that model’s tracks more often.
ACE Studio says the Flash version, a more lightweight open‑source release, is slated for late 2026.
No, the flagship remains closed‑source with access only via the ACE website, though the ACE‑Step 1.5 models are open‑source and can be run locally.
Discussion 0