Latest / 6 Ps in AI Pods (AI6P) / Say what? Adding audio voiceovers with AI, ethically 🗣️

Transcript

Say what? Adding audio voiceovers with AI, ethically 🗣️ · 6 Ps in AI Pods (AI6P)

• 0:00
  SPEAKER_00: Hi everyone, this is Karen Smiley and welcome to 6 P's in AI Pods on Substack.
• 0:08
  SPEAKER_00: I'm experimenting with adding audio voiceover like this on my articles.
• 0:12
  SPEAKER_00: This one is me, not AI.
• 0:15
  SPEAKER_00: Let me know if you find it useful or if you have any suggestions for improving it.
• 0:21
  SPEAKER_00: Say what?
• 0:23
  SPEAKER_00: Adding audio voiceovers with AI ethically.
• 0:28
  SPEAKER_00: Comparison of features and pricing on three AI-based voice cloning tools,
• 0:32
  SPEAKER_00: plus next steps in my experiment with AI-generated audio voiceovers for my Substack publications.
• 0:39
  SPEAKER_00: Thinking of adding audio voiceovers for your Substack posts?
• 0:42
  SPEAKER_00: Me too.
• 0:44
  SPEAKER_00: Wondering if generative AI tools can help you with your audio voiceovers?
• 0:48
  SPEAKER_00: Me too.
• 0:50
  SPEAKER_00: This article summarizes what I've learned in the past week about three AI-based
• 0:54
  SPEAKER_00: voice cloning tools,
• 0:56
  SPEAKER_00: my planned next steps for evaluating them,
• 0:58
  SPEAKER_00: and how you can participate.
• 1:01
  SPEAKER_00: I use two abbreviations in this article.
• 1:03
  SPEAKER_00: P-V-C, Professional Voice Cloning.
• 1:06
  SPEAKER_00: It's an 11 Labs feature that's also supported by WonderCraft.
• 1:10
  SPEAKER_00: And T-T-S, Text-to-Speech.
• 1:15
  SPEAKER_00: Background.
• 1:16
  SPEAKER_00: Substack has been abuzz lately about the benefits of adding audio and video for
• 1:20
  SPEAKER_00: subscribers and newsletter publishers.
• 1:22
  SPEAKER_00: I decided to explore adding audio voiceover to my Substack posts.
• 1:26
  SPEAKER_00: It needs to be,
• 1:27
  SPEAKER_00: one,
• 1:28
  SPEAKER_00: time efficient to save more of my weekly effort budget for actual writing,
• 1:33
  SPEAKER_00: and two,
• 1:33
  SPEAKER_00: minimal cost because I'm on a shoestring budget.
• 1:37
  SPEAKER_00: Using a canned, pre-trained generative AI voice can be free for a few short recordings.
• 1:43
  SPEAKER_00: Some links are in the end notes.
• 1:45
  SPEAKER_00: And it could be efficient.
• 1:46
  SPEAKER_00: It doesn't appeal to me, though.
• 1:48
  SPEAKER_00: I want my own voice saying my own words.
• 1:52
  SPEAKER_00: That leaves three options to consider.
• 1:55
  SPEAKER_00: One, manually recording my own voice with conventional tools, no AI.
• 2:00
  SPEAKER_00: No cost, but likely to be time-consuming.
• 2:03
  SPEAKER_00: Links for tips on doing this well are in the end notes.
• 2:07
  SPEAKER_00: Two, using an instant AI-based clone of my voice to generate audio.
• 2:12
  SPEAKER_00: And three, using a high-quality professional AI-based clone of my voice to generate audio.
• 2:18
  SPEAKER_00: Options two and three motivated me to tackle research into AI-based text-to-speech,
• 2:24
  SPEAKER_00: TTS,
• 2:25
  SPEAKER_00: voice cloning tools in the past few weeks.
• 2:28
  SPEAKER_00: As mentioned in my April retrospective, I'm now starting an experiment.
• 2:33
  SPEAKER_00: The experiment goal is to evaluate ethical AI-based voice cloning tools,
• 2:38
  SPEAKER_00: initially Speechify and Eleven Labs,
• 2:41
  SPEAKER_00: compare cost,
• 2:42
  SPEAKER_00: quality,
• 2:43
  SPEAKER_00: and effort versus manual recording for audio voiceovers,
• 2:47
  SPEAKER_00: and come to a decision,
• 2:48
  SPEAKER_00: probably in June.
• 2:50
  SPEAKER_00: My April 28th 6Ps and AI Pods article summarized AI-based voice cloning vocal cover
• 2:57
  SPEAKER_00: tools and identified which ones appear to be ethical.
• 3:00
  SPEAKER_00: Spoiler, most do not.
• 3:03
  SPEAKER_00: That article has details on ethics of 11Labs and Speechify.
• 3:08
  SPEAKER_00: On Tuesday, I learned about WonderCraft.ai voice cloning and started looking into its ethics.
• 3:15
  SPEAKER_00: WonderCraft.ai, a quick look at ethics.
• 3:18
  SPEAKER_00: WonderCraft's FAQ offers an 11Labs PVC promo code and links to the 11Labs voice lab
• 3:25
  SPEAKER_00: for professional cloning.
• 3:28
  SPEAKER_00: However,
• 3:28
  SPEAKER_00: WonderCraft also lets you configure your own voice model on their site with about
• 3:33
  SPEAKER_00: four minutes of spoken audio input.
• 3:35
  SPEAKER_00: They even provide a sample script.
• 3:38
  SPEAKER_00: The FAQ doesn't mention any guardrails to try to ensure that users only create a
• 3:43
  SPEAKER_00: custom non-PVC voice model in WonderCraft with their voice,
• 3:47
  SPEAKER_00: not an unauthorized voice.
• 3:49
  SPEAKER_00: WonderCraft only provides a page with a stern warning about legality.
• 3:53
  SPEAKER_00: Eye roll.
• 3:55
  SPEAKER_00: That's a potential ethical concern.
• 3:58
  SPEAKER_00: I'm also not yet finding any documents describing,
• 4:00
  SPEAKER_00: one,
• 4:01
  SPEAKER_00: the sources of the pre-built voices they offer,
• 4:04
  SPEAKER_00: or two,
• 4:04
  SPEAKER_00: how they protect your voice model from reuse for other purposes.
• 4:08
  SPEAKER_00: In fairness, I haven't yet read through all of their many policy PDFs.
• 4:14
  SPEAKER_00: Bottom line,
• 4:15
  SPEAKER_00: although WonderCraft may not be as strong ethically as I hope,
• 4:18
  SPEAKER_00: for now I'm adding it to my shortlist along with Speechify and Eleven Labs.
• 4:24
  SPEAKER_00: TTS Features and Pricing Models
• 4:28
  SPEAKER_00: In all three tools,
• 4:30
  SPEAKER_00: training a model for TTS with your own voice requires a monthly plan which is not free.
• 4:35
  SPEAKER_00: Higher quality voice models, PVC, cost more.
• 4:40
  SPEAKER_00: All three tool providers offer discounts for annual plans.
• 4:45
  SPEAKER_00: Pricing for all plans scales rapidly based on the number of minutes of audio recording needed.
• 4:51
  SPEAKER_00: To keep my costs down for option two and option three,
• 4:55
  SPEAKER_00: I'll need to set a minimal minutes budget that I can live with.
• 4:58
  SPEAKER_00: Caveat, this analysis isn't a broad evaluation of these three TTS tools for all needs.
• 5:05
  SPEAKER_00: Each tool includes other features, e.g.
• 5:07
  SPEAKER_00: multiple voices or seats, which are not relevant to my current needs for audio voiceovers.
• 5:13
  SPEAKER_00: Features I'm ignoring may be of value to people with other needs, so their evaluations may differ.
• 5:20
  SPEAKER_00: Speechify voice cloning.
• 5:23
  SPEAKER_00: For TTS, Speechify Limited is free and Speechify Premium is $139 a year.
• 5:28
  SPEAKER_00: However, neither supports cloning and using your own voice, and the focus seems to be personal use.
• 5:35
  SPEAKER_00: TTS plans and prices are summarized in the endnotes for Speechify Studio,
• 5:40
  SPEAKER_00: which includes additional capabilities beyond basic TTS.
• 5:44
  SPEAKER_00: Discounts for annual plans are substantial, over 50%.
• 5:49
  SPEAKER_00: Based on the length of time needed to train the voice model,
• 5:52
  SPEAKER_00: quote,
• 5:53
  SPEAKER_00: in seconds,
• 5:54
  SPEAKER_00: Speechify Professional's voice cloning seems comparable to instant non-PVC offerings.
• 6:02
  SPEAKER_00: Eleven Labs voice cloning.
• 6:05
  SPEAKER_00: Eleven Labs TTS plans and prices for basic and professional quality voice cloning
• 6:10
  SPEAKER_00: are summarized in the end notes.
• 6:12
  SPEAKER_00: Thank you to James Garside for sharing his example of a canned voice from his Eleven Labs trial.
• 6:19
  SPEAKER_00: Wondercraft Voice Cloning.
• 6:21
  SPEAKER_00: Wondercraft plans and pricing for basic and professional quality voice cloning are
• 6:25
  SPEAKER_00: summarized in the end notes.
• 6:27
  SPEAKER_00: It's not clear if the pro fee includes the 11 Labs Creator subscription for PVC at
• 6:35
  SPEAKER_00: $18.33 a month,
• 6:37
  SPEAKER_00: or if the promo code offsets the cost,
• 6:41
  SPEAKER_00: either once or recurring.
• 6:46
  SPEAKER_00: Tool Comparisons.
• 6:48
  SPEAKER_00: All cost and minute allocations are per month,
• 6:51
  SPEAKER_00: monthly prices assume a yearly plan,
• 6:54
  SPEAKER_00: and all plans considered allow commercial use of the generated audio recordings.
• 7:00
  SPEAKER_00: For basic cloning of one voice,
• 7:03
  SPEAKER_00: the table in the article summarizes comparisons of these three tools for basic
• 7:06
  SPEAKER_00: voice cloning.
• 7:08
  SPEAKER_00: Tool advantages vary by audio recording time limits, 30, 60, 120, or more minutes.
• 7:15
  SPEAKER_00: Details are in the endnotes.
• 7:19
  SPEAKER_00: Professional cloning for one voice.
• 7:22
  SPEAKER_00: A key pricing factor is whether the Wondercraft Pro account subsidizes or covers
• 7:27
  SPEAKER_00: use of the 11 Labs Voice Lab for PVC,
• 7:30
  SPEAKER_00: e.g.
• 7:30
  SPEAKER_00: via a promo code.
• 7:33
  SPEAKER_00: If not,
• 7:33
  SPEAKER_00: then the total Wondercraft Pro cost at all levels will be at least $18.33 more for
• 7:39
  SPEAKER_00: at least one month to build the model and import it into Wondercraft.
• 7:45
  SPEAKER_00: It's not clear whether the PVC model
• 7:49
  SPEAKER_00: built in Eleven Labs can continue to be used in Wondercraft even if the Eleven Labs
• 7:54
  SPEAKER_00: creator account is no longer active.
• 7:58
  SPEAKER_00: If not, the additional $18.33 a month will boost the total Wondercraft Pro cost every month.
• 8:08
  SPEAKER_00: Assuming that Speechify Professional's voice cloning is instant quality,
• 8:11
  SPEAKER_00: not professional quality,
• 8:13
  SPEAKER_00: only Eleven Labs and Wondercraft are considered here.
• 8:18
  SPEAKER_00: The table below summarizes comparisons of these two tools for professional voice cloning.
• 8:23
  SPEAKER_00: As with basic voice cloning, advantages vary by audio recording time constraints .
• 8:34
  SPEAKER_00: Detailed commentary is in the endnotes.
• 8:38
  SPEAKER_00: TLDR Summary of Pricing and Feature Comparisons for Voice Cloning
• 8:43
  SPEAKER_00: Assuming that WonderCraft Pro covers the Eleven Labs Creator PVC fee,
• 8:48
  SPEAKER_00: the table below in the article shows the overall results and costs per month by
• 8:52
  SPEAKER_00: audio recording limits up to 500 minutes per month.
• 8:58
  SPEAKER_00: Bottom line,
• 9:00
  SPEAKER_00: for low-volume basic voice cloning up to 30 minutes a month,
• 9:04
  SPEAKER_00: Eleven Labs Starter is the clear choice at $4.17.
• 9:09
  SPEAKER_00: 11Labs Creator will get you up to 120 minutes a month for $18.33.
• 9:15
  SPEAKER_00: For high volume,
• 9:17
  SPEAKER_00: basic voice cloning up to 500 minutes a month,
• 9:20
  SPEAKER_00: Speechify Professional is a clear choice at $32.09.
• 9:22
  SPEAKER_00: For professional voice cloning,
• 9:28
  SPEAKER_00: Eleven Labs' Creator or Pro plans look like the best choices,
• 9:31
  SPEAKER_00: although Wondercraft Pro 150 has an edge on cost in the 121 to 150 minute window.
• 9:38
  SPEAKER_00: Depending on Wondercraft promo codes and discounts, it could also be competitive in other windows.
• 9:45
  SPEAKER_00: Limitations.
• 9:47
  SPEAKER_00: This evaluation is so far based on reading available online materials about each of the tools.
• 9:52
  SPEAKER_00: No hands-on experience yet.
• 9:56
  SPEAKER_00: Second, some of the tool providers may run promotions or referrals, which significantly reduce the cost.
• 10:01
  SPEAKER_00: I've heard that WonderCraft does.
• 10:04
  SPEAKER_00: This could change the trade-offs and outcomes.
• 10:07
  SPEAKER_00: Only standard published prices were used for this analysis.
• 10:11
  SPEAKER_00: So question,
• 10:12
  SPEAKER_00: do any of you who have worked with these tools either notice any considerations I
• 10:17
  SPEAKER_00: missed or see anything I've misinterpreted about the tool's voice cloning
• 10:22
  SPEAKER_00: capabilities or pricing models?
• 10:24
  SPEAKER_00: If so, please leave me a comment.
• 10:28
  SPEAKER_00: Conclusions.
• 10:30
  SPEAKER_00: My final evaluation will weigh the trade-offs among audio quality,
• 10:33
  SPEAKER_00: time,
• 10:34
  SPEAKER_00: limits,
• 10:34
  SPEAKER_00: effort,
• 10:35
  SPEAKER_00: and cost for options one and two,
• 10:38
  SPEAKER_00: and option three if audio quality is needed.
• 10:42
  SPEAKER_00: Option one, recording my own voice without AI, is my baseline.
• 10:46
  SPEAKER_00: It's free for me.
• 10:48
  SPEAKER_00: I already have a Yeti microphone, a pop filter, and Audacity.
• 10:52
  SPEAKER_00: And audio recording time per month is unlimited.
• 10:56
  SPEAKER_00: The main cost will be my time.
• 10:59
  SPEAKER_00: For Option 2,
• 11:00
  SPEAKER_00: the Eleven Labs Starter Account looks affordable and provides the minimum voice
• 11:04
  SPEAKER_00: cloning capability I'm looking for.
• 11:09
  SPEAKER_00: Evaluation Plan is Option 1.
• 11:11
  SPEAKER_00: Make a few sample recordings manually.
• 11:13
  SPEAKER_00: Measure effort and quality.
• 11:15
  SPEAKER_00: Option 2.
• 11:17
  SPEAKER_00: Evaluate Eleven Labs with a Starter Account and compare effort and quality.
• 11:23
  SPEAKER_00: Key Question.
• 11:25
  SPEAKER_00: Is basic voice cloning good enough to add value to my publications?
• 11:30
  SPEAKER_00: At this point,
• 11:31
  SPEAKER_00: I'm crossing my fingers that one,
• 11:34
  SPEAKER_00: 30 minutes a month will be enough for now,
• 11:37
  SPEAKER_00: and two,
• 11:38
  SPEAKER_00: the Live and Lab starter quality with basic voice cloning will be good enough and
• 11:44
  SPEAKER_00: save me enough time versus manual voiceover recordings to justify the small monthly
• 11:50
  SPEAKER_00: budget hit.
• 11:52
  SPEAKER_00: What's next?
• 11:54
  SPEAKER_00: Writers,
• 11:55
  SPEAKER_00: if you are using one of these tools now for Substack audio voiceovers or decide to
• 12:00
  SPEAKER_00: try one,
• 12:01
  SPEAKER_00: I'd love to join forces with you on this evaluation.
• 12:03
  SPEAKER_00: Please share your experience and lessons learned and let me know if you'd like to
• 12:07
  SPEAKER_00: guest post here about it.
• 12:09
  SPEAKER_00: Message me.
• 12:11
  SPEAKER_00: I'll post in a future article about the results of my experiments with 11 Lab Starter and other options.
• 12:17
  SPEAKER_00: Subscribe for free to be automatically notified when new posts are published and to
• 12:21
  SPEAKER_00: let me know this AI topic interests you.
• 12:25
  SPEAKER_00: Readers, what do you think about audio voiceovers?
• 12:29
  SPEAKER_00: Please see my poll on whether you would listen to audio voiceover on my Substack articles,
• 12:33
  SPEAKER_00: if it's short,
• 12:34
  SPEAKER_00: if it's long,
• 12:35
  SPEAKER_00: maybe,
• 12:35
  SPEAKER_00: or if you prefer reading,
• 12:37
  SPEAKER_00: and leave me a comment to let me know.
• 12:40
  SPEAKER_00: Thanks for listening.