VoiceBook Production Proposal

Db1f7d2e17690904435c71e9b581a431?s=47 Nick Marsh
November 26, 2017
310

VoiceBook Production Proposal

Short version of a proposal to fund a voice recognition focused children's story telling app at Lost My Name

Db1f7d2e17690904435c71e9b581a431?s=128

Nick Marsh

November 26, 2017
Tweet

Transcript

  1. 2.

    Contents Tech - Will it work? Product - What is

    the idea? Targets - What’s the business goal? Project - How will we ship? Summary and Sign off
  2. 3.

    Remember that in the past... Tech - Will it work?

    <- this was a blocker Product - What is the idea? <- this was confusing Targets - What’s the business goal? <- this was unresolved Project - How will we ship? <- this was unresolved Summary and Sign off
  3. 4.

    Why? So much why. • Does is give all adults

    creative super powers to make magical meaningful connections with a child? ◦ Yes it does. • Will it help make us the most admired creative company in the UK? ◦ Yes it will. • Does it bring together tech and storytelling in a brilliant way? ◦ By jove it does. • Is it in line with our brand value of surprising, fun, bold and exuberant. ◦ Fuck yes it is. • Of course we should do this, but we should do it smart and good.
  4. 6.

    Technical Iterations v1: Keyword Spotting v2: Grammar / Rules v3:

    Hybrid Approach Only listen to certain *key* words • Rhyming words trigger false positives • Speaking quickly is a common point of failure Constrain Phonemes into a grammar • Stumbling over a sentence requires the whole sentence being read again • CPU / Memory intensive for longer sentences Mix and match between grammar and keyword spotting • Unreliable creation syntax • Stumbling grammars were still a failure point • Still trying to do too much
  5. 7.

    Where we are today v4.0 - Tiered System Formally tested

    on 10 English speaking subjects, 6 with US accents 96% success rate on our simpler grammar modes Very confident in tech and our ability to use UX and content to overcome barriers. Tech is not the blocker any more.
  6. 8.

    Demo Tiered Recognition System Speech Building Blocks • Speech is

    Occurring • Syllable Counting • Offline Recognition • Online Recognition* • Silence** *Online Recognition is a last resort (not free, slow, unreliable) **Silence also includes, umms, errrs, coughing and sighing Language specific recognition modes • Phoneme • Syllable • Word • Couplet • Phrase • Line • Stanza • Fast everything • Fast keyword spotting • Quickly uttered sentence • Custom Grammar • Rejecting
  7. 10.
  8. 11.

    Voice is used as a novelty element No relationship between

    form and content Suitable for a marketing agency to show off their tech ideas The equivalent of a personalised book where you swap the name out for your kids name
  9. 12.

    The Creative Brief for VoiceBook Create a VoiceBook for adults

    to read to children aged 2-5 that celebrates the incredible range of the human voice - and the specific joy of reading aloud to children. The VoiceBook should be a hoot (a technical term) and make everyone that reads it want to read it again and again - and tell their friends about it. Simply put, reading this VoiceBook should make you better at reading all stories out loud. The VoiceBook must be deliverable by summer 2017.
  10. 13.

    The Creative Brief for VoiceBook Create a VoiceBook for adults

    to read to children aged 2-5 that celebrates the incredible range of the human voice - and the specific joy of reading aloud to children. The VoiceBook should be a hoot (a technical term) and make everyone that reads it want to read it again and again - and tell their friends about it. Simply put, reading this VoiceBook should make you better at reading all stories out loud. The VoiceBook must be deliverable by summer 2017.
  11. 14.

    Deliverable by summer 2017 To ensure we ship we need

    to make a very simple, modular story This will match the creative with the delivery methodology and overall be less story and art direction work to manage. Our benchmark story is Oi Frog! Demo
  12. 15.

    Example Story - The day I lost my voice I

    lost my voice once, who knows how? I searched, all day and night. Perhaps I moo, just like a cow? MOO, MOO, MOO, MOOOOOOOOO Hmmm. That doesn’t sound quite right. I lost my voice once, just like that. I searched, all day and night. Perhaps I meow, just like a cat? MEOW, MEOW, MEOW MEEEEEOOOW Hmmm. That doesn’t sound quite right. Repeat At last I found my voice, hooray! I had searched day and night. My voice knows lots of words to say: WIBBLE, WOBBLE, BLIM BLAM, TIGGLE-TAGGLE FLUBBER-BLUBBER They sound exactly right! Share your silliest noise! The day I lost my voice
  13. 16.

    Art direction We imagine the art direction will be graphical,

    bold, typographically oriented and low on character animation. This will allow the voice interaction to come to the fore … and enable us to actually ship this by the summer!
  14. 17.
  15. 18.
  16. 19.
  17. 20.
  18. 21.
  19. 22.

    Business model and marketing concept We will make VoiceBook a

    free app with an appropriate link to find out more about Lost My Name’s incredible books. We will promote VoiceBook to our existing customers mainly using non-paid channels and work hard to build a relationship with Apple and arrange strong PR. If time allows we’d like to ship with a feature in the app that will easily allow users to record themselves and share the recording. If VoiceBook delivers on our goals (coming next) then we’ll explore other monetization concepts later.
  20. 24.

    Commercial Goals Make us famous • Win 10 Awards Open

    a new channel • 500,000 downloads in 12 months • Achieve some viral growth through users sharing the app with others Deliver a memorable and high quality product • 25% of installed users will read a VoiceBook in its entirety • 10% of users will read VB end to end multiple times Make money and acquire new customers • Attribute 25,000 physical books sales to VoiceBook • 50% of these sales are to new customers Business case doc
  21. 25.

    Data: Blinkies Make us famous • 1 award nomination Open

    a new channel • 9,100 downloads to date Deliver a memorable and high quality product • 96% of installed users use a Blinkie • 96% of users use a Blinkie more than once (average sessions 1.1) • Average session duration 21 minutes Make money and acquire new customers • 4% click through to LMN site (through parental gate) • 496 visits • 32k video ads watched, 2,384 watched to completion
  22. 26.

    Data: Snow Fox Make us famous • Looks like it

    has been in the top 100 kids for a few months Open a new channel • Looks like 75,000 downloads - front page of US app store Time sensitive app Not tied to a brand