How We Pick

So why trust us at all?Fair question, and it deserves an answer in the same breath rather than a page away. We haven’t heard these units. What we’ve done instead: compiled every published specification and linked its source, computed the compatibility arithmetic so you can check it yourself, and priced every pick live today — or not at all. Reproducible beats authoritative when you have no authority. And you can catch us being wrong, which is more than you can do with an adjective.

What we actually do

1. We compile the published specifications.From the manufacturer’s own product page, spec sheet or manual. Not from a retailer listing — retailer spec fields are unreliable and we do not treat them as a source. Not from another review site.

2. We cite every one of them. Every spec table on this site has a source column, and every row links to the document we read that value from. This is enforced in our code, not by editorial discipline: a specification cannot be added to our data layer without a source URL, so a spec with no source cannot render.

3. We say when a manufacturer doesn’t publish something. “Not published” appears throughout this site. Audio-Technica doesn’t publish tonearm effective mass. ELAC doesn’t publish a frequency-response tolerance. Klipsch doesn’t publish a wattage for The Fives II. Yamaha doesn’t publish a continuous 4-ohm rating. Those gaps are information, and essentially nobody else prints them.

4. We do the arithmetic, and show it. Formula, inputs, the source of each input, and the result. If you disagree with our answer you can find exactly where we went wrong, which is the entire difference between an argument and an assertion.

5. We price everything live. Or not at all.

The arithmetic we use

Three calculations do most of the work on this site. All three are standard.

Cartridge/tonearm resonance

F = 1000 / (2π × √(M × C))

M is total effective mass in grams (arm effective mass + cartridge mass + fixings); C is dynamic compliance in µm/mN. Target: 8–12 Hz — above record warps and footfall, below the lowest musical content. Worked through here, including the fact that Ortofon does not publish the frequency their compliance figure was measured at, which changes the answer.

Amplifier power vs speaker sensitivity

SPL = Sensitivity(1 W/1 m) + 10·log₁₀(P) − 20·log₁₀(d)

With one correction first: sensitivity is published at 2.83 V, which equals 1 W only into 8 ohms. Convert with P = V²/R before comparing speakers of different impedances. Worked through here.

Quarter-wavelength, for acoustic absorption

λ/4 = c / (4f), where c ≈ 343 m/s

It explains why a 2-inch panel absorbs 12% at 125 Hz and 112% at 1 kHz. Worked through here.

How we handle prices

Every price on this site comes from Amazon’s API and is stamped with the date it was fetched — currently July 17, 2026.

If a price is more than 48 hours old, it disappears and the button says “Check price on Amazon” instead. That is automatic and structural, not a policy we remember to follow. Amazon’s operating agreement requires displayed prices to be current, and a stale number is both a policy violation and a lie to you.

We never write a price into an article. There is no number on this site that a human typed.

Our selection criteria, per category

These are specific enough that you could apply them yourself and reach our list. That is what “reproducible” means, and it is the claim.

Turntables

1. Currently purchasable on Amazon with a verified ASIN. No exceptions — see below.
2. A published tracking force, adjustable or specified.
3. Published wow and flutter, or an explicit note that the maker does not publish it.
4. A stated cartridge, and whether it is user-replaceable.
5. Whether a phono stage is included — and if not, we count that cost against the price.

Speakers

1. Published sensitivity with its measurement condition. A bare dB figure is not comparable and we say so.
2. Published impedance; minimum impedance where the maker gives it.
3. Frequency response with a tolerance. A range without ±dB is not a specification.
4. We separate power handling from recommended amplifier power. They are different specs.

Amplifiers and receivers

1. RMS power with the full condition — load, bandwidth, THD, channels driven.
2. Dynamic or “peak” power is never treated as a rating. We flag it where it is being passed off as one.
3. Published 4-ohm behaviour, or an explicit note that there is none.
4. Phono input and its type.
5. Pre-outs, because they decide whether the unit can ever be upgraded.

DACs

1. Published THD+N and signal-to-noise.
2. Inputs and outputs — which is what actually decides the purchase, since they all measure well.
3. Headphone output power at a stated impedance, if it has one.

Room treatment

1. Published per-frequency absorption coefficients, including 125 Hz.
2. NRC alone is not sufficient — it excludes the bass by definition.
3. Whether the data is lab-measured or calculated. Primacoustic marks the difference; we repeat it.
4. Material and thickness, because the physics is thickness-bound.

Scoring

Every product carries per-metric scores out of 10 and an overall figure. Those are our editorial judgement applied to the published data — they are not measurements, not customer ratings, and not aggregated from anywhere.

That is why you will never see star-rating rich snippets from this site in Google. Marking up an editorial score as if it were aggregated user reviews would be structured-data fabrication, so we do not emit it. We would rather lose the snippet.

Every roundup names something we would skip

A roundup where every product is great is an advertisement. So every one of ours identifies at least one product we would not buy, with the published reason.

Every product also gets a cons list. Not a token one — a real one, drawn from what the specs say. If we could not find a genuine drawback, that means we did not look hard enough.

What would change our mind

A pick gets dropped or replaced when:

• It stops being purchasable. A top pick you cannot buy is the one failure this model cannot survive.
• A manufacturer publishes a figure that contradicts our reasoning.
• A spec source disappears and we cannot re-source the value.
• A better product appears in the bracket on the criteria above.
• You tell us we are wrong and you are right.

Why some obvious products are missing

You will notice no Rega, no Technics, no Wharfedale, no Cambridge DacMagic. Those are well-regarded products and their absence is deliberate: we could not verify them as currently purchasable through the retailer we link to. We would rather leave a gap than write a recommendation with a buy button that cannot exist.

Similarly, the Sony PS-LX310BT and STR-DH190 appear with no specifications at all. Sony blocks automated access across every domain it operates — we got HTTP 403 from sony.com, its regional sites, and the official manual PDFs. Their manuals are mirrored on third-party sites, but a mirror is not the manufacturer. So we publish nothing rather than something we could not verify at source. That rule cost us something there, which is roughly the point of having it.

Updates and corrections

The “Last verified” date on each page changes only when a human has actually re-verified it. It is not bumped automatically at build time to look fresh. That date is a promise; we keep it or we do not print it.

Prices revalidate continuously and need no human. Roundups get a human re-check quarterly. Any pick that becomes unavailable is replaced or removed in the same cycle.

Found an error? Tell us — it is the most useful message we get. Corrections are fixed, dated and logged visibly per our editorial policy.

How we make money

Affiliate commission, disclosed on every page that carries a buy link. It costs you nothing extra and it does not influence what we pick — and the structural reason it cannot is this page: our criteria are published and reproducible, so if commission were steering the picks, you could catch it by applying the criteria yourself and getting a different list.

The full disclosure is here.