Standard text-to-speech engines often fail at long-form narration because they read text with relentless, robotic uniformity. Real human narrators vary their cadence, pause to breathe, soften incidental thoughts, and inflect their pitch across character dialogue. In Voicemaker, producing professional, human-grade voiceovers requires combining two distinct layers: global voice settings in the sidebar to shape narrator baseline personality, and inline SSML tags alongside Custom Pauses in the editor to direct timing, emphasis, and pronunciation.
This guide walks through configuring Voicemaker to eliminate machine monotony across audiobooks, podcasts, and narrative video content.
1. Engine Selection and Baseline Voice Setup
Before touching markup tags, establishing the right base acoustic profile is essential.
- Select the AI3 Neural Engine: Standard or legacy voice models lack the prosodic modeling required for natural speech flow. Always choose the AI3 voice engine.
- Identify Voices with the "E" Suffix: In Voicemaker, voices featuring an "E" suffix in their name (for example, Jony in English US under AI3) support expressive Voice Effects such as calm, sad, happy, or angry. Choosing a voice without this suffix limits you to basic speed and pitch adjustments without emotional coloring.
- Calibrate Global Voice Settings:
- Speed: The default speaking speed is often slightly hurried for narrative storytelling. Open the Voice Settings icon in the right sidebar and dial the speed down slightly (to a relaxed pace). A moderate reduction allows listeners to absorb descriptions without listener fatigue.
- Pitch: Lowering global pitch slightly creates a richer, authoritative narrator persona. Conversely, raising pitch slightly works well when setting up distinct character presets for dialogue.
- Volume: Set global volume to maintain consistent headroom across chapters (avoiding high-gain settings that introduce clipping in downstream audio editing).
2. Pacing and Breath Control: Pause SSML vs. Custom Pauses
The most immediate giveaway of synthetic audio is absent or poorly timed breathing pauses. Voicemaker provides two complementary ways to control silence: manual pause tags and automated punctuation-based pauses.
Manual Insertion with Pause SSML
For deliberate, dramatic pacing, use the Pause dropdown in the top menu bar or type pause tags directly into the text editor:
- Attribution Pauses (200ms – 400ms): Insert brief pauses between direct quotes and narrator attributions (e.g.,
"I don't believe you," <break time="300ms"/> she whispered.). - Scene Shifts and Chapter Beats (1000ms – 2000ms): Use longer pauses between major thoughts, scene transitions, or following dramatic story cliffhangers.
Automating Rhythm with the Custom Pauses Feature
Inserting break tags manually into a 5,000-word chapter is tedious and clutters the script. Instead, configure Voicemaker's built-in Custom Pause engine:
- Click the Settings gear icon in the top-right corner above the editor.
- Select the Pauses tab in the General Settings window and toggle ON Custom Pauses.
- Configure silence durations in seconds for specific punctuation marks:
- Full stop / Period (
.): Set to0.7s–0.9sfor a natural sentence conclusion. - Comma (
,): Set to0.25s–0.35sfor natural clause separation. - Question mark (
?) & Exclamation mark (!): Set to0.6s–0.8sto let rhetorical weight land.
- Full stop / Period (
- Enable Between-Paragraph Pauses: Ensure there is a blank line between manuscript paragraphs. Voicemaker will automatically insert your designated paragraph break (such as
1.5s), giving listeners natural cognitive breaks between narrative blocks. - Special Pause Triggers (
@and#): Map custom symbols like@(e.g.,0.5s) or#(e.g.,1.2s) within the Custom Pauses menu. Inserting these single characters into your text allows instant, surgically placed pauses without bloated XML syntax.
3. Dynamic Inflection: Emphasis, Pitch, and Speed Modulation
Monotonous volume and pitch flatten emotional resonance. Voicemaker allows selective modulation on individual words or phrases.
Highlighting Key Words with Emphasis SSML
- Highlight a target word or phrase in the editor and click the Emphasis dropdown.
- Strong Emphasis: Increases vocal weight and intensity. Apply this sparingly to pivotal verbs, narrative twists, or forceful dialogue outbursts. Overuse leads to a disjointed, staccato delivery.
- Reduced Emphasis: Lowers vocal intensity and volume. Use this for parenthetical remarks, inner monologues, or whispered secrets.
Applying Subtle Pitch and Speed Adjustments
Voicemaker's SSML pitch and speed controls are engineered to be subtle to prevent unnatural synthetic artifacts:
- Pitch Adjustments: Highlight a question or character line, click Pitch in the top toolbar, and adjust the slider. A slight rise at the end of an interrogative phrase creates convincing inquiry inflection. A slight drop deepens foreboding narration.
- Speed Adjustments: Highlight contemplative or tension-filled sentences and decrease the speed slider via the Speed dropdown. Moderate adjustments create palpable suspense without breaking voice clarity.
Applying Expressive Voice Effects to Dialogue
For voices with the "E" suffix:
- Ensure the global voice effect in the sidebar remains set to Default.
- Select the specific dialogue sentence spoken by a character.
- Click the Voice Effect button on the top toolbar and select the intended emotion (such as whisper, calm, or angry).
- When you generate speech, the narration retains its neutral storytelling baseline while dialogue shifts into the selected emotional texture.
4. Accurate Pronunciation with Say-As SSML
Audiobook narration often breaks down when text-to-speech engines encounter dates, phone numbers, measurements, or abbreviations. Voicemaker's Say-As suite provides deterministic pronunciation control:
- Say-As Date & Time: Prevents ambiguous dates like
04/05/1920from being read as mathematical equations or inverted day-month formats. - Say-As Cardinal & Ordinal: Directs numbers to read either as counting amounts (
12-> "twelve") or rankings (12-> "twelfth"). This is crucial for historical texts ("the 14th century") and chapter numbers. - Say-As Spell-Out / Characters: Forces the engine to voice each letter individually (
say-as interpret-as="characters"). Use this for secret radio codes, initials, acronyms, or flight designations that the engine might otherwise attempt to pronounce phonetically as a single word. - Say-As Fraction & Unit: Ensures values like
1/2or15kmread properly as "one half" and "fifteen kilometers," preventing abrupt stumbles.
5. Production Workflow: Chunking, Audio Export, and Credit Budgeting
Delivering a multi-chapter audiobook or lengthy commercial voiceover requires an efficient production workflow:
Chunking Your Script
Never paste an entire 10,000-word manuscript into the editor at once.
- Break your text into scene-based or thematic chunks between 500 and 2,500 characters.
- Chunking makes fine-tuning SSML tags straightforward, allows rapid testing of difficult names or terms, and limits credit consumption if a line requires re-generation.
Export Settings for Professional Mastering
Above the Convert to Speech button, click Change next to Audio Settings:
- Output Format:
- MP3: Ideal for podcasts, standard web narration, or quick preview drafts.
- WAV: Essential for audiobooks. Choose uncompressed WAV to retain maximum dynamic range for mastering in digital audio workstations (DAWs) like Audacity, Reaper, or Pro Tools.
- Sample Rate:
- Select 44100 Hz (CD quality) for standard streaming distribution.
- Select 48000 Hz (Studio quality) when preparing audio for video synchronization or platforms demanding strict studio submission criteria like ACX and Audible.
Credit Management and Licensing
- Voicemaker calculates plan limits based on Converts, not downloads. Every click of "Convert to Speech" expends credits according to text length (note that CJK characters cost 2 credits each).
- Free accounts are restricted to 250 characters per conversion and do not include full commercial rights.
- Commercial narration for audiobooks, client voiceovers, and monetized media requires a paid tier such as Starter ($5/month), Creator ($10/month), Pro ($20/month), or dedicated Audiobook Creation Packs, which grant commercial licensing and higher character burst limits.
- Always preview tricky sentences in 50-character isolated snippets before converting whole sections to preserve your monthly credit allocation.
Summary Checklist for Natural Narration
- Pick the Right Voice: Choose an AI3 neural voice carrying the "E" suffix for emotional flexibility.
- Establish the Baseline: Set speed slightly slower and pitch slightly deeper in the sidebar Voice Settings.
- Turn On Custom Pauses: Map commas to ~0.3s, periods to ~0.8s, and blank lines to ~1.5s in Pause Settings.
- Direct the Key Moments: Use Emphasis SSML for impactful phrases and Say-As tags for dates, fractions, and acronyms.
- Export Studio-Ready Files: Download scene-based chunks in 48000 Hz WAV format for final assembly and mastering.