Hands-on first look

Suno Speech (beta) review: narration and a score in one take, fine print still missing

Oct 2, 2026 · Verdict provisional while the beta evolves. This review's song is on the way.

Verdict (provisional)

Suno Speech turns a script, poem, or loose idea into spoken audio with original background music, generated together as one track (Suno blog). It went into public beta for everyone on Oct 1, 2026, on web, iOS, and Android (Suno release notes).

Each generation cost 10 credits for two takes, the same as a song, and most finished in about 10 seconds (our test). The controls are thin.

The fine print is still missing. Suno hasn't published a length cap or said whether its commercial-rights terms cover Speech. In our tests, the gender switch worked blind with a neutral Tone but seemed to lose out to the Tone text in earlier runs, and one take kept talking after its script ended.

Provisional take: cheap and quick to try if you already pay for Suno. Don't build a podcast or audiobook workflow on it until length, rights, and listening results are in. This verdict will change.

• Who it's for: Suno users making scored intros, gifts, bedtime stories, and spoken-word pieces.
• Who should skip it (for now): anyone who needs a specific voice, line-by-line fixes, or written commercial terms. Try ElevenLabs.

Key facts

What it is, and what's new

In the web app, Speech (marked BETA) is a tab on Create, next to Songs and Sounds. Simple mode has one prompt box. Advanced mode adds a Script field, a Tone field ("Describe the delivery: tone, pacing, mood and setting"), Vocal Gender (Male/Female), Background music (Off/On), and a Variety slider labeled Off, Normal, High, Extra, and Max; Normal is the default (our test).

That matches The Verge's description, except The Verge mentions a "speech style" setting. We found no style or voice picker; that direction goes in Tone.

Hands-on test

What we measured: cost, takes, speed, durations, and the interface. Listening so far is one person's ear: Nathan Crowe, our editor, gave his impressions of 3 takes, then did a blind gender check on 4 takes. No scores yet.

Setup: suno.com web app, tested Oct 2, 2026 on a Pro (Annual) account.

What came out (every row cost 10 credits for 2 takes):

Measured findings

• Cost: 100 credits for 10 generations (80 for T1–T7, 20 for the gender check). A flat 10 per generation, whether takes ran 0:12 or 2:19. No cost was shown before generating.
• Speed: most generations took about 10 seconds; T7 took about 15 and T3 about 20.
• Reliability: one submit failed ("Failed to fetch"); the retry was charged once.
• Consistency: takes of one script varied a lot in length (T1: 0:17 vs 0:25; T3: 1:28 vs 2:19; gender check A: 0:13 vs 0:38). We haven't checked whether short takes drop words.
• First listen (Nathan only, 3 takes, qualitative): Nathan listened to the T1 0:17, T3 1:28, and T6 0:34 takes. To his ear, the accent stayed consistent within each take. That's a small point against Suno's own accent-drift warning, but only for takes of 1:28 or less. T1's delivery felt stilted and its accent a little odd. T3 and T6 sounded almost the same: a British female voice, but good. No scores yet.
• Gender check (blind, Nathan only, 4 takes): same short script, Tone "Calm, neutral American narrator, medium pace.", music off, Variety Normal. Run A was set to Male, Run B to Female. Nathan identified all four takes correctly, blind: both A takes as an American male, both B takes as an American female. All four were clear and evenly paced, with no artifacts or mispronunciations he could hear. With a neutral Tone, the gender setting worked in our test.
• Tone vs. gender (observation): the earlier T1, T3, and T6 takes were set to Male but sounded female or ambiguous. Their Tone descriptions were written for the podcast, accent-test, and trailer scripts. So in our tests, the Tone text appeared to outweigh the gender switch. Those runs also used music and different scripts, so we can't isolate the cause.
• Overrun (one instance): one Male gender-check take (0:38) read the full script correctly, then kept going with extra, unscripted audio. Its sibling ran 0:13. That's one possible cause of the length variance above, alongside dropped or rushed lines, which we haven't ruled out.

Not tested yet: scored listening against ElevenLabs, music/voice balance, script fidelity, the length cap, and export and download behavior. We'll update this review as we run them.

Prompts used

All scripts are original, with no real people, brands, or artist names. Each of T1–T6 also had a short delivery-and-music description in the Tone field; we didn't record that text word for word, so it isn't quoted here. We also didn't record the Variety setting for T1–T7 (Normal is the default). The gender check is quoted in full.

T1. Podcast intro (Advanced mode; also run as T1b with music off)

Script: Welcome to Song Tested. Every week we take one AI music tool, run it through the same seven tests, and tell you what it made, what you can download, what you own, and what it costs. No hype, no affiliate-driven verdicts. Today: a tool that talks. Let's see if it should.
Settings: Vocal Gender Male; Background music On (T1), Off (T1b)

T2. Audiobook-style narration (Advanced mode)

Script: The ferry ran twice a day, and Marta had missed the first one on purpose. She sat on the seawall with her bag between her boots and watched the gulls work the bait shop roof. Her brother would be waiting on the other side, or he wouldn't. Either way, the tide didn't care. At four o'clock the horn sounded across the water, low and patient, and she stood up before she'd decided to.
Settings: Vocal Gender Female; Background music On

T3. Accent stress test (Advanced mode)

The T2 script pasted three times (1,148 characters).

Settings: Vocal Gender Male; Background music On

T4. 30-second ad read (Advanced mode)

Script: Your coffee shouldn't taste like a meeting. Harrow Hill Roasters ships small-batch beans the week they're roasted, straight to your door. No subscription maze, no mystery blends. Just good coffee from people who answer their own email. Use code TESTED for ten percent off your first bag. Harrow Hill. Wake up on purpose.
Settings: Vocal Gender Female; Background music On

(Harrow Hill Roasters is fictional, made up for this test.)

T5. Spoken word (Advanced mode)

Script: Rent's a rumor in a borrowed coat. I count the stairs, not the years I owe. Window cracked so the street can talk. Every corner keeps a key I lost. Keep it low, keep it moving. Nothing here was ever proven.
Settings: Vocal Gender Male; Background music On

T6. Trailer narration (Advanced mode)

Script: Nobody remembers who built the bridge. They only remember the night it held. One town. One river. One choice nobody wanted to make. This winter, the water rises.
Settings: Vocal Gender Male; Background music On

T7. Simple mode (no script)

A tired night-shift nurse giving herself a quiet pep talk in a parking garage at 6 a.m., over slow, warm piano

Gender check (Advanced mode; Run A Male, Run B Female)

Script: This is a short voice test. I am reading three plain sentences at an even pace. Thanks for listening.
Tone: Calm, neutral American narrator, medium pace.
Settings: Background music Off; Variety Normal; Vocal Gender Male (A), Female (B)

Pricing (checked Oct 2, 2026)

What Speech costs: 10 credits per generation for two takes (our test), the same as a song (Suno v6 FAQ). Suno's pricing page doesn't mention Speech.

Suno's general plan lineup:

Downloads are capped at 20/mo on Pro and 60/mo on Premier (Suno downloads FAQ). Suno hasn't said whether Speech downloads count against that cap, and we haven't tested downloads yet.

Rights: Suno says "songs downloaded while subscribed are granted commercial use rights" (Suno help, edited Sep 3, 2026). Free-plan songs are non-commercial (help). Neither page mentions Speech.

Pros and cons (provisional)

Pros

• One prompt gets you voice and score in a single track (Suno blog).
• Same price as a song, and fast (our test).

Cons

• Suno hasn't published a length cap, download-cap rules, or rights terms for Speech.
• No cost shown before generating (our test).
• Thin voice control: Male/Female plus a free-text box, no voice picker (our test), and no advertised cloning (AlphaSignal). In our tests, Tone text appeared to outweigh the gender switch: earlier takes set to Male sounded female or ambiguous.
• Takes of the same script varied widely in length, and one ran past its script with unscripted audio (our test; one instance).
• Suno warns that accents drift and pauses run long (Suno blog).
• No API announced.

Alternatives

ElevenLabs generates narration and music separately and arranges them on a timeline.

ElevenLabs list prices checked Oct 2, 2026 (a promo was running).

FAQ

Is Suno Speech free? Suno says the beta is open to everyone (Suno blog). It hasn't published whether free accounts get the same access or limits; we tested on Pro.

Can I get just the voice, without music? Yes. Advanced mode has a Background music Off/On switch (our test).

Sources

All checked Oct 2, 2026 (CT).

1. Suno, "Introducing Speech (beta)," Jack Brody, Oct 1, 2026
2. Suno release notes, "Introducing Speech (beta)," Oct 1, 2026
3. Suno release notes index
4. Suno pricing
5. Suno help, "What rights do I have with a paid subscription?" (edited Sep 3, 2026)
6. Suno help, Rights & Ownership category
7. Suno v6 FAQ
8. Suno downloads FAQ
9. The Verge, "AI music maker Suno now generates spoken words," Jess Weatherbed, Oct 2, 2026
10. AlphaSignal, "Suno Speech generates narration and music in one pass"
11. ElevenLabs pricing
12. ElevenLabs, ElevenCreative Studio docs
13. Music Business Worldwide, UMG and Sony's second suit against Suno (Sep 2026)
14. Song Tested hands-on test of Suno Speech (beta): suno.com web app, Oct 2, 2026, Pro (Annual) account