ElevenLabs Prompts

Act as my audio production director. I am publishing an audiobook of [BOOK_TITLE], a [GENRE] book of about [WORD_COUNT] words aimed at [TARGET_AUDIENCE]. In ElevenLabs Projects, create a new project and choose a narrator voice that fits the tone: [VOICE_GENDER], [AGE_RANGE], with a [NARRATION_STYLE] delivery (options like warm and conversational, deep and dramatic, light and witty). Set stability to [STABILITY_VALUE] (higher for narration consistency, around 60-75) and similarity boost to [SIMILARITY_VALUE] (around 70-80 to keep the voice true to the sample). Split the manuscript into chapters of [WORDS_PER_CHAPTER] words, assign the narrator voice to the main narration and a second contrasting voice for [SECONDARY_CHARACTER_OR_QUOTE_ROLE], and use voice settings overrides only for dialogue lines so the narration stays uniform. Generate chapter by chapter, checking pacing after each, then export the final MP3. Output format: a numbered checklist of the exact Projects setup steps with the voice names, stability and similarity values, and chapter split plan filled in for my book.

Act as a voice branding consultant. I want to clone my own voice in ElevenLabs so every piece of branded audio sounds like me, for the use case of [USE_CASE] (for example weekly podcast intros, course lessons, or ad reads for [BRAND_NAME]). Walk me through the Voice Cloning workflow: record a clean sample of [SAMPLE_LENGTH_MINUTES] minutes in a quiet room, reading [SAMPLE_SCRIPT_TYPE] (conversational script with varied sentences, not monotone), uploaded in [AUDIO_FORMAT]. After cloning, test the voice with three sample sentences covering [EMOTION_1], [EMOTION_2], and [EMOTION_3], and adjust stability (start at [STABILITY_START], lower toward [STABILITY_END] if the delivery sounds too flat, raise if it sounds unstable) and the style exaggeration slider ([STYLE_VALUE]) until the read matches my natural cadence. Then create [NUMBER_OF_ASSETS] ready scripts for [CONTENT_TYPE] in a [TONE] tone, each under [MAX_WORDS] words. Output format: a step-by-step cloning checklist plus the test sentences and the final scripts, clearly labeled.

Act as a conversation designer for ElevenLabs Conversational AI. Design a voice agent for [BUSINESS_NAME], a [BUSINESS_TYPE] in [CITY_REGION], whose job is to [AGENT_GOAL] (for example book appointments, answer FAQs, or qualify leads). Define the agent persona: name [AGENT_NAME], voice style [VOICE_STYLE], speaking pace [PACING], and tone [TONE]. Write the opening greeting (under 30 words), the five most common caller questions with short natural answers (each under 60 words), and the fallback response when the agent does not know an answer: acknowledge honestly and offer [ESCALATION_ACTION]. Set the first response latency expectation, choose [LANGUAGE] from the supported 32+ languages, and add conversation guardrails: never discuss [FORBIDDEN_TOPICS], always confirm [CONFIRMATION_DETAIL] before finalizing. Output format: persona card, greeting script, Q&A table with question, answer, and follow-up, plus guardrail list.

Act as a foley artist using ElevenLabs Sound Effects. I need layered sound effects for a [SCENE_TYPE] scene in my [PROJECT_TYPE] (short film, ad, game, or YouTube video). The scene lasts [DURATION_SECONDS] seconds and takes place in [SETTING]. Generate [NUMBER_OF_LAYERS] separate effect layers: [LAYER_1] (for example footsteps on gravel), [LAYER_2] (ambient room tone), [LAYER_3] (a key action sound like a door slam), and [LAYER_4] (optional texture like wind or rain). For each layer, write the exact text prompt to paste into the Sound Effects generator, specifying material, distance, and intensity, for example 'heavy wooden door slamming shut, close perspective, echoey hallway'. Include notes on which layers to fade in or out and the final mix order. Output format: a numbered layer list with the exact generator prompt, duration, and mix notes per layer.

Act as a dubbing producer using ElevenLabs Dubbing Studio. I have a [VIDEO_LENGTH] minute video in [SOURCE_LANGUAGE] about [VIDEO_TOPIC] aimed at [TARGET_AUDIENCE], and I want a professional dub in [TARGET_LANGUAGE] plus optionally [SECOND_TARGET_LANGUAGE]. Plan the full dubbing workflow: upload the source video, enable speaker detection for [NUMBER_OF_SPEAKERS] speakers, and map each detected speaker to a voice that matches their [GENDER] and approximate age [AGE_RANGE]. Set the translation tone to [TONE] (formal or casual), choose whether to keep the original background audio ([KEEP_BACKGROUND_YES_NO]) for music and ambience, and review the auto-generated transcript segment by segment for names and terms that must stay untranslated: [TERMS_TO_KEEP]. After generation, do a timing QA pass so no line overruns its on-screen speaker. Output format: a phase-by-phase dubbing plan with speaker mapping table, transcript review checklist, and QA steps.

Act as a podcast producer. I run a [PODCAST_GENRE] podcast called [PODCAST_NAME] for [TARGET_AUDIENCE], and each episode is [EPISODE_LENGTH] minutes. Build a repeatable ElevenLabs text-to-speech pipeline: pick [HOST_VOICE_COUNT] voices for hosts [HOST_NAMES] with contrasting timbres ([TIMBRE_A] and [TIMBRE_B]), set stability around [STABILITY] for natural delivery and similarity boost at [SIMILARITY]. For each episode, I will supply a script of [SCRIPT_WORD_COUNT] words with speaker labels and [AD_READ_COUNT] ad reads marked for a slightly more energetic delivery (lower stability by 10 points on those sections). Include a pre-generation checklist (script proofread, phonetic spellings for names like [TRICKY_NAME_1] and [TRICKY_NAME_2], numbers written out), and a post-generation QA step comparing the first 60 seconds against the previous episode for voice consistency. Output format: the pipeline as a numbered SOP with voice settings, script formatting rules, and QA checkpoints.

Act as a character voice director using ElevenLabs Voice Design. I am creating an original character voice for [PROJECT_NAME], a [PROJECT_TYPE] (game, animation, audio drama). The character is [CHARACTER_NAME], a [AGE] year old [ROLE] with personality traits [TRAIT_1], [TRAIT_2], and [TRAIT_3]. In Voice Design, describe the voice with these sliders in mind: gender [GENDER], age [AGE_BUCKET] (young, middle-aged, old), accent [ACCENT], and descriptive text like '[VOICE_DESCRIPTOR_1]' and '[VOICE_DESCRIPTOR_2]' (for example gravelly, warm, mischievous). Generate [NUMBER_OF_VARIANTS] variants, test each with the line '[TEST_LINE]' in [EMOTION] and '[TEST_LINE_2]' in a contrasting emotion, and pick the winner based on [SELECTION_CRITERIA]. Then lock the design, save it to my Voice Library as [LIBRARY_NAME], and write [NUMBER_OF_LINES] sample dialogue lines in the character's speech pattern for [SCENE_CONTEXT]. Output format: design parameters, test results table, and the final dialogue lines.

Act as an operations consultant. My team at [COMPANY_NAME] holds [MEETING_TYPE] meetings of about [DURATION] minutes in [LANGUAGE], and I want to use ElevenLabs Speech to Text to turn recordings into usable notes. Design the workflow: record in [RECORDING_SETUP], upload the audio, run transcription with speaker diarization enabled for [EXPECTED_SPEAKERS] speakers, and export the transcript as [EXPORT_FORMAT]. Then define the post-processing pass: identify action items and assign each to [OWNER_ROLE], flag decisions with timestamps, and summarize the meeting in [SUMMARY_LENGTH] bullet points for [STAKEHOLDER_AUDIENCE]. Add a quality rule: spot-check [SPOT_CHECK_MINUTES] minutes of audio against the transcript for accuracy, especially names like [PROPER_NOUN_1] and technical terms like [TERM_1]. Output format: the workflow as a numbered SOP with the summary template and action-item table format.

Act as an instructional designer. I am building an e-learning course called [COURSE_TITLE] about [COURSE_TOPIC] for [LEARNER_AUDIENCE], with [MODULE_COUNT] modules averaging [MINUTES_PER_MODULE] minutes each. In ElevenLabs, choose [NARRATOR_COUNT] narrator voices: a primary instructor voice ([INSTRUCTOR_STYLE]) and a secondary voice for examples and quotes ([SECONDARY_STYLE]). Set stability to [STABILITY] and similarity boost to [SIMILARITY], and use the pronunciation dictionary to lock correct readings of [TERM_1], [TERM_2], and [TERM_3]. Format each module script with section headers, pause markers of [PAUSE_LENGTH] seconds between concepts, and a slightly slower pace for the [COMPLEX_TOPIC] module. Include a consistency QA step: compare module 1 and module [MODULE_COUNT] side by side for voice drift. Output format: voice casting sheet, script formatting rules, pronunciation entries, and the QA checklist.

Act as a direct-response creative director. I am running [PLATFORM] ads for [PRODUCT_NAME], a [PRODUCT_CATEGORY] targeting [TARGET_AUDIENCE], and I want to A/B test voiceovers in ElevenLabs before spending on media. Write [NUMBER_OF_VARIANTS] ad scripts of [WORD_COUNT] words each, all selling [CORE_BENEFIT] but with different angles: [ANGLE_1] (problem/agitate), [ANGLE_2] (social proof), and [ANGLE_3] (offer/urgency). For each variant, specify the voice ([GENDER], [AGE_RANGE], [DELIVERY_STYLE]), stability [STABILITY], and similarity boost [SIMILARITY], plus a pronunciation note for [BRAND_NAME_PRONUNCIATION]. Define the test: run each variant with [BUDGET_PER_VARIANT] budget for [TEST_DAYS] days, and declare a winner on [WINNING_METRIC] with a minimum [MINIMUM_LIFT] percent lift. Output format: the three scripts with voice settings each, then the test plan table.