Speech writing help: structure, voice and delivery notes

Why writing for the ear is a different skill

A speech and an essay can cover the same material and still need to be written completely differently, because a listener and a reader take in language in fundamentally different ways. A reader who loses the thread of a long, complex sentence can stop, go back, and re-read it. A listener cannot. Once a sentence has been spoken, it is gone, and if it was too dense to follow the first time, the listener is now behind and possibly missing whatever comes next while trying to catch up.

That single constraint changes almost everything about how a speech should be built. Sentences need to be shorter and more direct than they would be on the page, because a listening audience needs to be able to parse each one in real time, without re-reading. Rhythm and repetition, which can feel heavy-handed in a written essay, become useful tools when spoken aloud: a phrase repeated at intervals gives a listening audience something to hold onto and signals that a point matters. Signposting, saying "first," "second," and "finally" out loud, or explicitly saying "there are three things I want to leave you with," would feel clunky and over-explained in a written piece, but it does real work in a spoken one, because it helps a listener track structure they cannot see on a page.

This also changes how a good speech reads on the page before it is ever spoken. A speech that looks slightly repetitive or slightly plain in a text document can still work extremely well out loud, because the qualities that make prose feel sophisticated on paper (varied sentence length stacked with subordinate clauses, understated word choice, sentences that reward a second reading) are often exactly the qualities that make a passage hard to follow the first and only time an audience hears it. Judging a speech draft the way you would judge an essay draft is one of the more common ways a genuinely good speech gets edited into a weaker one. The only reliable test is reading it aloud, at the pace it will actually be delivered, and noticing where the listener (even if that listener is only you, reading to yourself) loses the thread.

The common speech formats and their shared bones

This guide covers speech writing at a structural level across the formats that come up most often: a wedding toast or other celebratory speech, a graduation or commencement-style speech, a business or conference keynote, a eulogy, and an award acceptance speech. Each of these has its own conventions and expectations. A eulogy carries a different emotional register than a keynote, and a wedding toast has different length norms than a commencement address. But underneath those differences, speeches that actually work tend to share the same four bones.

A wedding toast and a eulogy will fill in these four bones very differently in tone and content, but both fail in the same way when they skip one of them: an opening that wastes the audience's initial attention on housekeeping, a speech that wanders between several unconnected points instead of supporting one throughline, a string of generic sentiments with no specific stories attached, or an ending that just runs out rather than arriving somewhere.

How each format tends to differ in practice

The four shared bones stay constant, but how they get filled in shifts noticeably from one occasion to the next. Knowing what a given format tends to expect helps avoid a speech that feels tonally off for the room, even if it is well written on its own terms.

A wedding toast is short, warm, and usually funny in places, but it needs at least one moment of real sincerity to avoid feeling like a comedy set. It typically centers on one or two specific stories about the couple, addresses both people if the speaker knows both, and closes with an actual toast line rather than trailing into applause.

A graduation or commencement-style speech speaks to a room full of people at a shared transition point, which means it works best when it acknowledges something true and specific about that particular group or moment rather than reciting universal advice that could apply to any graduating class anywhere. A commencement speech built entirely from generic wisdom, however well phrased, tends to blur into every other commencement speech a listener has heard.

A business or conference keynote usually needs a clearer argument or a single big idea than a personal or celebratory speech does, since the audience is there for information or a point of view, not just an occasion. It still benefits enormously from concrete stories and examples rather than a slide-deck's worth of abstractions read aloud, but the throughline tends to be a claim or insight rather than a personal narrative.

A eulogy carries the heaviest emotional weight of any of these formats and the most risk in getting the tone wrong. It benefits from specific, human memories rather than a formal recitation of accomplishments, and from acknowledging the real range of emotion in the room rather than staying relentlessly upbeat or relentlessly solemn throughout. A eulogy built from a handful of small, true stories tends to land better than one that tries to summarize an entire life in general terms.

An award acceptance speech is usually short, sometimes strictly timed, and needs to balance genuine gratitude with something more than a list of names. Naming people matters, but a list of thank-yous without any specific reason attached to each name reads as an obligation being fulfilled rather than gratitude being expressed. One specific detail per name, even a brief one, does more than a longer, generic list.

Matching the speaker's actual voice

The fastest way to make a speech fall flat is to write something generic and insert the speaker's name into it. An audience that knows the speaker will notice immediately if the speech doesn't sound like something that person would actually say, and an audience that doesn't know the speaker will still sense a generic quality even if they can't name why. A speech that sounds like it could have been given by anyone, about anyone, has failed at its most basic job.

Matching a real voice requires gathering real material from the speaker: actual phrases they use, specific stories only they would tell, their sense of humor, how formal or informal they naturally are, what they'd never say even if it fit the moment. Writing a strong speech in someone else's voice is fundamentally a listening exercise before it is a writing one. The more specific and personal the source material, the more the finished speech will sound like the speaker rather than like a template with a name dropped in.

In practice, this usually means the person providing the instructions is either the speaker themselves or someone who knows the speaker closely enough to supply real detail on their behalf, such as a maid of honor writing on someone else's approval, or a colleague gathering material for a boss's keynote. Either way, the same rule applies: the more the source material sounds like the actual person, in their actual words, the more the finished speech will too. A request that only says "write me a wedding speech" with no further detail leaves a specialist to write something competent but generic, since there is nothing specific to build from.

Delivery notes worth including

ApplicationWriters is not a speech-coaching or delivery-training service, and a written speech can't replace practicing it out loud. That said, a few practical notes built into the written speech itself genuinely help a speaker deliver it well:

• Natural pause points marked where a breath or a beat helps the material land, especially after a punchline or before a serious turn.
• Words or short phrases worth emphasizing, so a speaker reading it for the first time has a sense of where the weight falls.
• A realistic estimate of speaking time based on the word count, generally around 130 to 150 words per minute for a measured, unhurried pace. This is a general guideline, not an exact promise, and every speaker's natural pace differs, so it is worth testing against your own read-aloud speed before an event with a strict time slot.

Roughly matching word count to speaking time by occasion helps set expectations before writing even starts. These are general conventions, not fixed rules, and any actual time slot given by an event should take priority:

These ranges assume a measured, unhurried pace with room for pauses. A speaker who talks quickly or who plans to move through material briskly may fit more words into the same time, and one who pauses often for effect will need fewer. Reading a draft aloud at the pace intended for the actual event is the only reliable way to confirm timing before the day.

An example: from generic to specific on the opening line

The first version could open almost any event and gives the audience nothing to hold onto. The second earns attention in two sentences, sets up a specific story, and tells the audience there's a real reason for what follows, which is exactly what an opening needs to do before the housekeeping and thank-yous, if any, come later.

What to gather before you start

A speech is only as personal as the material it's built from, so the most useful thing a client can do before starting an order is gather real, specific detail rather than trying to write polished sentences first. Useful raw material includes: two or three specific stories or memories connected to the occasion, actual phrases or expressions the speaker uses regularly, a sense of the speaker's natural tone (formal, dry, sentimental, blunt), names and relationships of anyone who should be mentioned along with a specific reason for mentioning them, and any hard constraints such as a strict time limit or topics that should be avoided.

None of this needs to arrive organized or in speech form. A rough voice memo, a list of bullet points, or even a scattered set of memories in no particular order is enough for a specialist to work from. What matters far more than polish is specificity: a single vivid, true detail is worth more to the finished speech than several paragraphs of general sentiment.

It also helps to say upfront what should not be in the speech. Family history is rarely uncomplicated, workplaces have politics, and a wedding party sometimes includes an ex or a rocky relationship that shouldn't be referenced even in passing. A short note about topics to avoid, names to leave out, or jokes that would land wrong with a particular relative saves far more editing time later than leaving it for a revision request after a first draft has already gone in a direction that needed correcting.

Common mistakes that undercut a speech

• A speech that reads well on paper but is exhausting to listen to, because the sentences are too long and complex to follow by ear.
• Cramming in too many stories or points, so nothing gets the space to actually land with real weight.
• An ending that just stops, rather than building toward a clear, memorable close.
• A speech that doesn't sound like the person delivering it, because it was written generically instead of built from their real voice and stories.
• Opening with a long list of thank-yous before saying anything the audience actually wants to hear.
• Ignoring the actual time slot and writing a speech that runs noticeably long or short once actually read aloud at a natural pace.
• Using vocabulary or sentence patterns that suit a written document but sound stiff or unnatural when spoken, such as heavy use of semicolons, parentheticals, or long qualifying clauses.

How ApplicationWriters helps with a speech

Choose Create from my information and a specialist writes your speech from the stories, phrases and details you provide, so it sounds like you rather than a generic template with your name inserted. If you already have a draft, choose Improve my draft to sharpen structure and voice while keeping your own material, or Edit and proofread for a lighter pass. Every order is priced from the length you set, shown before you pay. Whichever engagement type you choose, the goal is the same: a speech that sounds like the person delivering it, built for how it will actually be heard rather than how it will look on a page.

Frequently asked questions

Get speech writing help