# Why Ptichi exists: voice and speech for everyday life

Free exercises for speech, voice and reading aloud. Find what gets in the way, try one useful change and listen for the difference. Desktop app in development.

[Canonical page](https://ptichi.com/product/about)

Ptichi helps you name a speech or voice problem, practise one useful change and hear whether it helped.

Ptichi is being built as a free desktop app for recording, feedback, repeated practice and local history. Recordings and core analysis stay on your device by default.

Ptichi helps people speak more clearly and explore their voice through practical exercises, tools and guided practice.

## Practice with what you have

You can start without recording. Say a short phrase, change one thing and try it again in your own words. Our published exercises and the English text helper are free to use without an account.

Recording adds another option: listen back and compare attempts. Use your usual recorder today; this website does not capture audio. The desktop project is being built separately for local recording and comparison.

## Why Ptichi listens before it coaches

A lot of speech advice starts with the correction.

Slow down. Add energy. Use more pitch. Pause more. Sound confident.

The problem is that a correction can be perfectly sensible in general and still be wrong for the attempt you just made.

Ptichi is being built around the opposite order:

**say it → listen → choose one listener-relevant target → change one thing → compare.**

That order matters because the first job is not to make the recording look more impressive. It is to find out what actually needs changing.

The [first-recording design note](/research/why-the-first-screen-starts-with-one-recording) follows that idea into the developing desktop screen: how one recording leads to listening, and why a weekly activity count cannot tell you how much you learned. It is a source review, not a released app or a usability result.

If the point was already clear, adding extra emphasis may only make the delivery busier. If the final condition disappeared, a global instruction to “speak with more confidence” is decorative. If the microphone moved between takes, the fuller sound may belong to the setup rather than the speaker.

So Ptichi starts with the attempt. The cue comes later.

That is a less dramatic product demo than an instant score. It is also a more useful way to practise.

## Take A is supposed to be ordinary

The first take is not an audition.

You do not need to perform your best possible voice before the exercise has even started. Take A exists to capture what happens when you say the message in your current, ordinary way.

This is surprisingly important. People often improve the baseline as soon as they know it is being recorded. They sit straighter, speak slower, lean towards the microphone and add deliberate emphasis. Then Take B arrives with very little left to change.

For practice, that is not a disaster. It is just bad experimental housekeeping.

A useful baseline is not the worst version of you and not the polished version of you. It is a reasonably natural attempt under a repeatable setup.

If pressing Record changes everything, make a couple of neutral recordings first. Let the microphone become boring. Then start the comparison.

## One cue is not a limitation. It is the point.

Real speech contains dozens of things that software could show you: timing, pauses, energy, pitch, repetitions, sentence length, microphone level and much more.

Showing all of them at once does not automatically create a better training tool. Sometimes it creates a very expensive way to become self-conscious about eight unrelated graphs.

Ptichi's current design principle is narrower: **one listener target, one bounded change.**

For example:

> The launch is still planned for Friday, but legal approval is not complete.

You might decide that the condition after “but” is the part that must survive.

Take A: say it normally.

Cue: make the condition easier to recover.

Take B: change one thing — perhaps the boundary, the prominence of one key word or the amount of background before the point.

Then listen for that target.

If Take B is louder, lower, slower, more energetic and more theatrical at the same time, you may like it more. You also have no idea which change helped.

The one-change rule is deliberately boring. Boring is useful when you want to learn something from the comparison.

## Why Ptichi asks for changed wording after Take B

A second recording can improve for a very simple reason: you have just said the same sentence once already.

That is not fake progress. Repetition is part of practice. But it is not the same thing as being able to use the change on a new sentence.

This is why the Ptichi product model separates **same-material acquisition** from **changed-material transfer**.

After A and B, the useful next question is:

> Can I keep the same communication idea when I stop copying the sentence?

Suppose your target was making a condition easier to hear. Put the original wording away and explain another conditional message in your own words.

If the change survives, that is more interesting than Take B alone.

If it disappears, that is useful too. You may have learned the sentence rather than the skill.

A later attempt with less feedback is harder again. Ptichi does not collapse all of these into one “voice score” because they answer different questions.

## Why the program does not need to fill the screen with metrics

Metrics are useful when they answer a question you already have.

They are less useful when the software measures something because it can and then asks you to care about it.

A pause length can be measured. That does not mean every pause needs correcting. Pitch movement can be measured. That does not mean a wider range is automatically more natural. Energy can be measured. That does not mean louder speech is clearer speech.

Ptichi's product rules therefore put **meaning before metric**.

The intended listener effect comes first. Automatic evidence may support that work when the capture and analysis are trustworthy. When a capability is not trustworthy enough, it should stay unavailable rather than producing a confident-looking number anyway.

A missing metric is better than a fake one.

## What a desktop Ptichi session is being built to do

The desktop application is still in development; there is no public installer today. The current product design is intentionally small enough to explain without a feature bingo card.

A typical session is being built around this sequence:

1. **Microphone ready.** Select or recover microphone access and make sure the capture is usable.
2. **Take A.** Record a short attempt without trying to satisfy a wall of metrics.
3. **Replay.** Listen before receiving the main cue.
4. **One bounded change.** Work on a specific listener-relevant target.
5. **Take B.** Record again under comparable conditions.
6. **Compare.** Judge A against B for the target you chose rather than for a global vibe.
7. **Changed wording.** Try the same communication skill on new words.
8. **Local control.** Save, reopen, export or delete under the local-first data boundary.

The design names several product objects around that flow.

### Listening Loop

The Listening Loop is the core A/B interaction: listener target → Take A → self-listen → one cue → Take B → comparison → changed-wording transfer.

The important feature is not that the software can play two audio files. Any recorder can do that. The product question is whether the loop helps you make a useful decision about what to change next.

### Microphone Passport

The Microphone Passport is being designed as **capture truth**, not a microphone beauty contest.

Its job is to help answer questions such as: is the selected input usable, did the relevant capture condition change, and which measurements can be trusted for this take?

It should not produce a universal “microphone quality: 84/100” badge. A cheap microphone in a stable setup may be perfectly useful for a practice task. A premium microphone with changing processing can make A/B attribution worse.

### Intervention Card

An Intervention Card connects a communication target to a bounded cue, the evidence needed to support it, the main overcorrection risk and a transfer task.

In plain English: Ptichi should know **why** it is giving you the instruction, what would count as useful evidence, and when the advice should stop.

“Use more pitch” is not enough. “Make the corrected date easier to hear; do not turn every word into emphasis; test it again on new wording” is closer to a trainable intervention.

### Learning State

Ptichi's design keeps several outcomes separate: control, transfer, self-monitoring, effort and later retention.

That prevents a common product trick where one attractive number silently stands in for everything.

A better Take B may show useful control in this session. It does not automatically show that you will remember the change tomorrow, use it in a meeting or need less feedback next time.

Those are different questions. The product should behave as if they are different questions.

## The feature we want most is less dependence on the feature

A useful rehearsal product should eventually get out of the way.

If Ptichi helps you notice a boundary, you should later be able to make that boundary without staring at a graph. If the microphone helps you hear a contrast, the long-term target is not to need the microphone before every sentence. If a cue helps once, later practice should test whether you can reproduce the effect with less cueing.

This is why “more engagement” is not automatically a product win here.

The best session may end with you closing Ptichi and having the real conversation.

## What we are building

Ptichi is growing into guided practice programmes and installable tools. The current desktop targets are macOS on Apple Silicon and Windows 11 x64. Public installers are not available yet. A development target is not a compatibility guarantee or a release date.

Current public tools and materials are free. This is our present development phase, not a promise that every future service will always be free. Free access also does not mean the application source is open source.

## What guides the work

A useful exercise serves something you want another person to hear or understand. Your voice is not a personality test, and your accent is not a defect to score. We prefer one understandable change to a mysterious overall rating.

A more deliberate delivery is not automatically better if it feels uncomfortable. Practice should not require forcing your voice. Ptichi is not a diagnosis or treatment service.

Sources, our interpretation and actual project observations are labelled separately. A working interface or a better second attempt does not establish lasting learning; listener judgments, new wording and later use need their own evidence.

## Who is behind it

Ptichi is maintained by **Metalhatscats**. “We” is the project voice, not a claim that a research institute or a large coaching team is behind the site.

[Try a short exercise](/practice/boundary-contrast), read [why recorded voice can feel unfamiliar](/problems/why-your-recorded-voice-sounds-unfamiliar), or see [what is available and what is being developed](/product/roadmap). You do not need to wait for the app to begin.

Interactive controls remain on the canonical page. This is a saved text representation, not a completed practice or a product result.

```json
{
  "@context": "https://schema.org",
  "@type": "WebPage",
  "@id": "https://ptichi.com/product/about#webpage",
  "breadcrumb": {
    "@type": "BreadcrumbList",
    "@id": "https://ptichi.com/product/about#breadcrumbs",
    "itemListElement": [
      {
        "@type": "ListItem",
        "position": 1,
        "name": "Ptichi",
        "item": "https://ptichi.com/"
      },
      {
        "@type": "ListItem",
        "position": 2,
        "name": "Product",
        "item": "https://ptichi.com/product"
      },
      {
        "@type": "ListItem",
        "position": 3,
        "name": "Why Ptichi exists: voice and speech for everyday life",
        "item": "https://ptichi.com/product/about"
      }
    ]
  },
  "headline": "Why Ptichi exists: voice and speech for everyday life",
  "description": "Free exercises for speech, voice and reading aloud. Find what gets in the way, try one useful change and listen for the difference. Desktop app in development.",
  "url": "https://ptichi.com/product/about",
  "inLanguage": "en",
  "dateModified": "2026-10-21",
  "creditText": "Ptichi editorial",
  "publishingPrinciples": "https://ptichi.com/product/editorial-policy",
  "isPartOf": {
    "@id": "https://ptichi.com/#website"
  }
}
```