0:19
Explore
Shobhit Banga - Drop 3/15: Introducing the Speech Recognition Benchmark on @voicearen
VoiceArena · Nov 2026 · 2:14 · AI-annotated
- Length
- 2:14
- Scenes
- 26 · 5.2 s avg
- Product on screen
- 1:22
- Features shown
- 5
- Call to action
- 1:59 · 89%
- Voiceover
- Yes
00:00:00 / 02:14:02
VoiceArena’s 2:14 launch film runs 26 scenes, about 5.2 s each. The product shows up at 1:22 and 5 features are revealed. The call to action lands at 1:59 (89% in), over a voiceover. The video alternates a locked-off talking head in a sunlit office with clean indigo-on-white motion graphics, structuring the pitch as four numbered chapters (1/4 to 4/4) that each pair a live-action intro with a pop-in card demo, then close on a product reveal card, dataset roadmap, and shield logo. A traveling glowing dot, blur-to-focus title snaps, and one-by-one card reveals keep the moderate pace lively and readable.
Footage neededEstimated needs: 2 capture groups — your screens or recordings · 1 official logo — reused across the film · 18 filmed passages — shoot, supply or replace · 2 graphic passages — draw or supply
Is this your launch? to your site or press kit.
More like this
0:35
0:32
Caveman 4.0 + Caveman Cloud: why use many token when few do trick (YC F26)
Caveman · B2B · Oct 2026
Shobhit Banga - Drop 3/15: Introducing the Speech Recognition Benchmark on @voicearen, scene by scene
· 26 scenes as text
01 · Hook · 0:00–0:06
A man in a white t-shirt sits in a bright office in a locked-off medium shot, speaking to camera and gesturing with both hands. A blue lower-third with his name and title slides in at the lower right. Large windows and an iMac sit behind him.
On-screen text: “Shobhit Banga”, “Co-Founder & CEO”, “Voice Arena”
Camera: Static
02 · Hook · 0:06–0:09
A man in a white t-shirt talks and gestures in a bright office with whiteboards and a large window, held in a static chest-up shot. Two lines of white text appear in the top left, and a blue shield logo with a waveform mark sits in the top right corner.
On-screen text: “Dreaming up”, “an ASR benchmark”
Camera: Static
03 · Feature reveal · 0:09–0:15
A man in a white t-shirt sits in a bright office with tall windows, speaking to camera and gesturing with both hands. A white caption block reads “1/4 / Every use case” at the left edge, and a small blue shield logo sits in the top right. The locked-off frame holds steady while he talks.
On-screen text: “1/4”, “Every”, “use case”
Camera: Static
04 · Feature reveal · 0:15–0:18
A man in a white t-shirt talks and gestures with both hands in an office, whiteboards and a window behind him. Small white sparkles and a glowing ringed orb float near his hands. White text on the left states the headline and subtitle.
On-screen text: “1/4”, “Every use case”, “Many datasets, one benchmark”
Techniques: Particles
Camera: Static
05 · Feature reveal · 0:18–0:31
A light-blue slide titled 'Every use case of speech recognition' builds a grid of eight flat icon cards. A glowing dot hops to each corner and the card pops in one by one, from Multi-speaker to 'and more'.
On-screen text: “Every use case of speech recognition”, “Multi-speaker”, “Overlapping voices”, “Call centre”, “Phone audio at 8 kHz”, “8 kHz”, “Entities”, “Digits, passwords, names”, “4 5 2 1”, “p@ssw0rd', 'Emily Carter', 'Dictation', 'Speak it, see it typed', 'Child speech', 'Young voices', 'Long-form', 'Hours, not seconds', '1:24:06', 'Far-field', '1:”
Camera: Static
06 · Feature reveal · 0:31–0:32
A man in a white t-shirt talks and gestures in a bright office, framed from the waist up. White text reading 2/4 and Every language fades in on the left. A small blue shield logo with a waveform holds the top right corner.
On-screen text: “2/4”, “Every language”
Camera: Static
07 · Feature reveal · 0:32–0:37
A man in a white T-shirt speaks to camera in a bright office, gesturing with both hands, while a dotted world map with moving light pulses overlays the left side. The heading “Every language” and subline “For billions of people” stay locked at the left edge.
On-screen text: “2/4”, “Every language”, “For billions of people”
Techniques: Particles
Camera: Static
08 · Feature reveal · 0:37–0:39
A man in a white shirt speaks and gestures in a sunlit room with windows behind him. On the left, a dotted world map twinkles beneath a static headline and subline, with a small blue shield logo in the top right corner.
On-screen text: “2/4”, “Every language”, “For billions of people”
Techniques: Particles
Camera: Static
09 · Feature reveal · 0:39–0:41
A man in a white shirt gestures mid-speech, framed in profile against a sunlit window. On the left, a dotted world map twinkles beneath locked-off white text. A blue shield logo sits top right.
On-screen text: “2/4”, “Every language”, “For billions of people”, “100+ languages”
Techniques: Particles
Camera: Static
10 · Feature reveal · 0:41–0:45
A man in a white t-shirt sits in a bright office with tall windows, talking to the camera and gesturing with both hands. A small step counter and the headline 'Public and private sets' fade in at the top left, while a blue shield badge stays fixed in the top right corner.
On-screen text: “3 / 4”, “Public and private sets”
Camera: Static
11 · Feature reveal · 0:45–0:56
A locked-off light layout with the indigo headline above two white cards. The left Private card holds three blurred audio rows; the right Public card fills with three readable transcript rows, then short captions fade in under each card.
On-screen text: “One test set, split in two”, “Audit it yourself”, “Check its accuracy and relevance”, “Kept out of training data”, “Lowers the chance of contamination”
Camera: Static
12 · Feature reveal · 0:56–1:01
A man in a white t-shirt sits centered in a bright office, speaking to camera and gesturing with both hands. White text reading 4/4 and Metrics per use case holds on the left, and a blue shield logo sits in the top right corner. The frame stays locked off.
On-screen text: “4/4”, “Metrics per use case”
Camera: Static
13 · Feature reveal · 1:01–1:14
A white slide titled 'Real errors count. Valid spellings don't.' stacks two cards. The top compares a reference and model line, flagging one wrong word with a 20% metric. The bottom draws a Hindi word-path diagram ending in a checkmark and a 0% metric, while a glowing dot pulses across the elements.
On-screen text: “Real errors count. Valid spellings don't.”, “Word error rate”, “20%”, “1 wrong word out of 5”, “OIWER”, “The model wrote “अंग्रेजी”.”, “It matches a valid path, so no error.”, ““I know English” in Hindi. All three spellings are valid, so all count as correct.”, “0%”
Camera: Static
14 · Feature reveal · 1:14–1:22
A white slide titled “Different use cases, different metrics” fades in, then three bordered cards slide up in sequence, each pairing a metric name with a short example. A small shield logo sits top right, and an indigo pill reading “Beyond word error rate” pops in under the first card.
On-screen text: “Different use cases, different metrics”, “1”, “Word error rate”, “Words wrong out of words said”, “turn on the chicken lights”, “Said “kitchen”. One real error.”, “Dictation, captions”, “2”, “OIWER”, “Every valid spelling counts', 'अँग्रेजी', 'अंग्रेजी', 'अँग्रेजी', 'Three valid spellings, all correct', 'Languages with many ways to write', '3', 'Task success率”
Camera: Static
15 · Product intro · 1:22–1:29
A man in a white t-shirt stands in a bright office with grid windows, gesturing with both hands while speaking. A checkmarked list builds on the left, then the headline “Taking the first step” fades in below it. A blue shield mark sits in the top right corner, and the shot fades to white.
On-screen text: “Every use case”, “100+ languages”, “Public and private sets”, “Metrics per use case”, “Taking the first step”
Camera: Static
16 · Product intro · 1:29–1:34
A pale lavender card centers the VoiceArena.com lockup above a dark indigo headline that snaps from a soft blur into sharp focus. A dark blue “Live today” pill with a white dot then pops in below, ringed by a faint halo.
On-screen text: “VoiceArena .com”, “Speech Recognition Benchmark”, “Live today”
Camera: Static
17 · Feature reveal · 1:34–1:42
A white title card assembles in the upper left: a Dataset 1 pill, a large indigo Monsoon title, and a subtitle, then a rounded panel with two speaker waveforms as a thin playhead line sweeps left to right. A dark indigo pill caption pops in below the panel.
On-screen text: “VoiceArena .com”, “Speech Recognition”, “Benchmark”, “• Live today”, “Dataset 1”, “Monsoon”, “Conversational ASR Test Set”, “Conversational speech in 5 languages”
Techniques: Kinetic type
Camera: Static
18 · Feature reveal · 1:42–1:49
A man in a white T-shirt sits in a bright windowed office, gesturing as he talks to camera. On the left, six dataset entries build into a vertical timeline, a glowing dot travels down the line, and the caption More languages. More use cases. appears below.
On-screen text: “Dataset 1”, “Live”, “Dataset 2”, “Next few days”, “Dataset 3”, “Next few days”, “Dataset 4”, “Next few days”, “Dataset 5”, “Next few days”
Techniques: Kinetic type
Camera: Static
19 · Other · 1:49–1:53
A man in a white shirt talks and gestures in a bright office, whiteboards and a leafy window behind him. On the left, the words Independent. then Third-party. fade in one at a time in bold white type.
On-screen text: “Independent.”, “Third-party.”
Techniques: Kinetic type
Camera: Static
20 · Other · 1:53–1:55
A man in a white shirt gestures while talking in a bright office. Static white text on the left reads Independent. Third-party. and a blue shield logo sits in the top right corner.
On-screen text: “Independent.”, “Third-party.”
Camera: Static
21 · Other · 1:55–1:59
A man in a white shirt sits in a bright office with whiteboards and glass doors, gesturing with both hands as he speaks. The camera holds a locked-off medium shot while a blue shield logo sits in the top right and the white text overlays fade out at the end.
On-screen text: “Independent.”, “Third-party.”, “Trusted by academia”, “and industry”
Camera: Static
22 · CTA · 1:59–2:04
Locked-off shot of a man in a white t-shirt in an office, gesturing with both hands as he speaks to camera. At left, a white caption with a stacked-disc icon reads 'Public test sets on Hugging Face'; 'All 5 languages' fades in below. A blue shield logo sits top right.
On-screen text: “Public test sets on Hugging Face”, “All 5 languages”
Camera: Static
23 · CTA · 2:04–2:07
A man in a white t-shirt sits in a bright office, gesturing with both hands while speaking. Whiteboards and a window fill the background. A white headline block sits at left for the first half, then clears, while a blue shield logo holds in the top right corner.
On-screen text: “Public test sets on Hugging Face”, “All 5 languages”
Camera: Static
24 · CTA · 2:07–2:10
A man in a white t-shirt sits at a desk in a bright office, speaking to the camera and gesturing with both hands. Large windows and office chairs fill the background. A small blue shield logo with sound bars stays in the top right corner as he smiles.
Camera: Static
25 · Outro · 2:10–2:11
A washed-out figure fades to white; a blue dot glows at left, then a shield icon holding a sound wave and the text Voice Arena fade in at center.
On-screen text: “Voice Arena”
Camera: Static
26 · Outro · 2:11–2:14
A blue shield with a waveform sits left of the wordmark Voice Arena on a pale background, a soft halo fading around the shield.
On-screen text: “Voice Arena”
Camera: Static