Same input
Both apps receive the same article, document, scan, language sample, and account-region view.
Research protocol v0.1 · reviewed August 24, 2026
Text-to-speech app test protocol
A useful comparison needs the same files, device conditions, and measurements for every app. The fixture manifest and hashes are now frozen, and wave 01 records the first ReadGoodr observations plus Speechify's public US price displays. Incomplete rows stay visibly incomplete.
Core test rules
Recorded conditions
We log app version, iPhone model, iOS version, network, date, plan, and every manual step.
Observation before opinion
Timings, failures, OCR samples, and screenshots are stored separately from subjective voice preferences.
Reproducibility gate
What is frozen before a result can count
Fixture manifest
Record every URL and filename, a SHA-256 hash for each local fixture, its license or usage basis, and the human reference text used for OCR scoring.
Timing events
Start at the defined untouched state; stop at the first comprehensible spoken word. Run five cold and five warm trials per app, alternate app order, and record a 120-second timeout as a failure.
OCR calculation
Compare extracted text with a fixed human transcription and report word error rate and character error rate alongside omissions, inventions, and reading-order errors.
Price normalization
Capture US web and App Store checkout totals on the same date. Compare annual with annual and monthly with monthly; mark an unavailable period as not offered.
Evidence wave 01
First observations, including the failure
ReadGoodr was observed on an iPhone 16 Pro simulator running iOS 18.5 with bundle version 2. These are single-run functional observations, not speed or quality winners. Speechify's figures below are official public US web displays, not a completed checkout.
T02 · partial pass
Selectable-text PDF
A controlled 10-page PDF imported through Safari share. ReadGoodr showed 1/10, navigated to pages 2 and 10, displayed the correct page markers, and disabled next navigation at 10/10. Full extraction scoring and human audio review remain pending.
T03 · observed limitation
Image-only PDF
A controlled five-page image-only PDF reached ReadGoodr through the same share path, but the build returned “No text found” instead of invoking OCR. Separate camera and document-scan OCR still need their own physical run.
T05 · partial
Direct text and quota behavior
The exact 150-word fixture imported and reached the player. With all 30 Pro Voice hours already used, Rachel opened the extra-hours screen and the player then exposed robot fallback. Alabama–Darlene prepared and entered visible playback state. No human listening result is claimed.
T10 · public displays
Speechify price views
Speechify's official US web page displayed $29/month and $139/year for Premium. Its Free card advertised up to 1.5×, 10 robotic-sounding voices, and TTS-only features. Checkout, tax, renewal, and cancellation evidence remain pending.
Frozen fixture hashes: direct text
5ee04498…13205, selectable PDF bdddb47f…f128, image-only PDF f2739beb…134. Full SHA-256 values are retained in the run manifest.Test matrix
13 shared tests, no moving goalposts
Every row produces a timestamped evidence record for both apps. A row cannot be called complete from marketing copy alone.
| ID | Scenario | Controlled input | What we record | Status |
|---|---|---|---|---|
| T01 | Clean web article | The same public English-language article URL | Import success, cleanup quality, steps to play, and time to first comprehensible speech | Fixture defined · matched run pending |
| T02 | Text-based PDF | The same 10-page selectable-text PDF | Import success, reading order, page navigation, highlighting, and first-audio time | Wave 01 · ReadGoodr partial pass; Speechify pending |
| T03 | Scanned PDF | The same image-only five-page PDF | OCR success, omitted or invented words, reading order, and time until playback | Wave 01 · image-only PDF rejected; camera OCR pending |
| T04 | Photographed book page | The same curved, evenly lit printed page | Capture steps, OCR word error sample, paragraph order, and playback result | Fixture defined · matched run pending |
| T05 | Direct short-text import | The same 150-word sample pasted into each app | Number of user actions and time from untouched app state to first comprehensible speech | Wave 01 · ReadGoodr partial; audible check pending |
| T06 | Long document | The same 20,000-word document | Import reliability, processing time, resume behavior, and navigation | Fixture defined · matched run pending |
| T07 | English voice and speed | The same 500-word English passage | Available voices, intelligibility at 1×/2×/3×, and maximum selectable speed | Fixture defined · matched run pending |
| T08 | Swedish text | The same 500-word Swedish passage | Language detection, pronunciation notes, available voices, and highlighting | Fixture defined · matched run pending |
| T09 | Follow-along reading | The same article with paragraph and sentence markers | Highlight alignment, tap-to-seek behavior, scrolling, and recovery after manual scroll | Fixture defined · matched run pending |
| T10 | Price, trial, and cancellation | New US web and App Store account views for each available channel on the same date | Displayed checkout total, normalized billing period, trial terms, renewal disclosure, and cancellation path | Public web displays verified; matched checkout pending |
| T11 | Control Center copied text | The same 150-word clipboard sample launched from the actual ReadGoodr Control Center control | Whether text reaches the app, error state, number of actions, and time to first comprehensible speech | Fixture defined · matched run pending |
| T12 | Control Center copied URL | The same article URL on the clipboard, launched from the actual ReadGoodr Control Center control | Whether the URL reaches import, error state, cleanup result, number of actions, and time to first speech | Fixture defined · matched run pending |
| T13 | Limited mode, Trial, Pro, and Max boundaries | The same direct non-URL text, article URL, scan, and existing document under each entitlement state | Limited Apple-voice fallback after an active opt-out; URL, OCR, scan, and file gating before network, camera warmup, or library creation; Trial, Pro, and Max continuation; expired and billing-issue paywall behavior | Fixture defined · matched run pending |
Evidence record for every run
- Protocol version, fixture manifest, hashes, test date, app version, plan, device, iOS version, account region, and network.
- A screen recording from the untouched input through the first comprehensible spoken word.
- Five cold and five warm runs per timing scenario, with app order alternated and every timeout retained.
- The number of deliberate user actions, import failures, warnings, and recovery steps.
- OCR output checked against a human reference with WER, CER, omissions, inventions, and reading-order errors recorded.
- Visible price, billing period, trial, renewal text, and the path to manage or cancel the subscription.
Evidence and official sources
Official competitor sources establish advertised features and prices. The ReadGoodr wave entry records controlled physical observations. Neither replaces the still-pending matched runs. All were reviewed on August 24, 2026.
- ReadGoodr physical test wave 01
Controlled iPhone 16 Pro simulator observations for direct text, a selectable-text PDF, an image-only PDF, quota fallback, and visible playback state.
- Speechify pricing
Free and Premium plan details, the $29 monthly and $139 yearly US web displays, speed, voices, languages, scan, and cloud integrations.
- Speechify iOS product page
iOS features including highlighting, scan and listen, and broader AI tools.
- Speechify on the US App Store
Supported Apple platforms, public app description, and in-app-purchase disclosures.
Next
See the current decision guide
The comparison page separates verified public facts from launch assumptions and pending measurements, so you can decide whether either app fits today.
Compare ReadGoodr and Speechify