THE PRODUCT
Nari Qwen3-TTS and Qwen3-ASR
Dev-praised open TTS/ASR stack for speed and cost, but the Nari inference engine isn't open-sourced and voice-cloning fears dominate lay audiences.
THE VERDICT
REALITY SCORE · OUT OF 10 · CONFIDENCE LOW
COMPOSED FROM
SENTIMENT · 34 REVIEWS
OUR VERDICT
// Honest verdicts are the whole point. We only monetise products we'd actually recommend.
AT A GLANCE · QUOTABLE
- Rating: 5.8 / 10 (low confidence)
- User voices: 34 across 2 platforms
- Sentiment: 35% positive · 30% negative
- Updated: Sep 15, 2026
GYIBB rates the Nari Qwen3-TTS and Qwen3-ASR 5.8/10 based on 34 user voices from 2 platforms. Confidence: low. Source: https://gyibb.com/ai-voice/nari-qwen3-tts-and-qwen3-asr
BUY IF
Devs praise the speed-cost 'pareto frontier' vs closed APIs like GPT-Live-1
- + Qwen3-ASR reported leading HF's open ASR leaderboard
- + Active ecosystem: derivatives like Darwin TTS already built on Qwen3-TTS
- + Tech report detailed enough for users to improve their own Whisper inference
SKIP IF
Nari's inference engine repo is not open-sourced despite 'open source' framing
- − Voice-switching bug reported mid-way through a 33-second clip
- − No independent evals yet; claims unverified (INTERNET layer absent)
- − Strong misuse concerns: phone voice-authentication and identity-theft risks dominate general-audience reaction
Where the layers disagree ⚡
6 CONTRADICTIONS DETECTEDVIDEO title claims 'ElevenLabs just got nuked by open source,' but USER leaderboard comment reports ElevenLabs still leads its blind TTS comparison — only Darwin TTS (a Qwen3-TTS derivative) tops the open-model tier.
VIDEO and community framing say 'open source,' but USER comments plus a team reply confirm the Nari inference engine repo is NOT open-sourced — only the base Qwen3-ASR weights are.
USER (HackerNews) reality is utility-focused (speed, cost, Whisper improvements); VIDEO comment reality is risk-focused (10-second convincing clones, refusing voice authentication with banks/insurers) — same tech, opposite value judgments.
USER bug report of voice switching mid-way through a 33-second clip sits uneasily against the VIDEO's ElevenLabs-killer framing; long-form voice consistency is unproven in provided data.
USER comment explicitly demands independent evals to verify quality claims, and the INTERNET layer is missing — no third-party expert verification exists in this dataset.
ALIGNMENT: USER citations of Qwen3-ASR topping HF's open ASR leaderboard partially support the VIDEO's claim that open source now rivals closed APIs, at least on the ASR side.
WHERE THEY AGREE +
WHERE THEY DON'T −
Where the 34 sources came from
VIEW EVERY CITATION →The four realities of the Nari Qwen3-TTS and Qwen3-ASR
Most review sites collapse everything into one number. We keep the layers separate so you can see where reality bends.
What actual buyers say
What reviewers showed on camera
ElevenLabs just got nuked by open source
Jeff Geerling · 663,709 views
"[comment] So when you called me yesterday asking for $1000 that wasn't you? [comment] My insurance offered to use my voice as a way to authenticate over the phone. I said absolutely not. This is exactly why [comment] Don't ever use audio as…"
What the press said
What the brand says
no brand page found
SIMILAR IN THIS CATEGORY
See all →
DATA SOURCES & AUDIT
34 data points across 2 platforms, synthesized via GYIBB's Truth Engine and fact-checked against source data before publication.
CONFIDENCE: LOW · ANALYSED: SEPTEMBER 15, 2026 AT 09:38 AM · PROMPT V1.0 · READ METHODOLOGY →
Was this review helpful?
Embed this review
